Multilingual_Embeddings.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Multilingual_Embeddings.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "J-HL27XLAVFo",
        "12gYRQI9TLoy"
      ],
      "toc_visible": true,
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/ELehmann91/Thesis_Multilingual_Transferlearning/blob/master/Multilingual_Embeddings.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0nxMDHTuaNU-",
        "colab_type": "text"
      },
      "source": [
        "# Download"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kPmxTTHUsMQx",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "a6be6eee-ab02-430a-eb7d-12855f3b988e"
      },
      "source": [
        "from google.colab import drive, files\n",
        "import pickle\n",
        "\n",
        "drive.mount('/content/gdrive')\n",
        "path ='/content/gdrive/My Drive/Thesis_ecb_ecoicop'"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i-ZknJaRrjQX",
        "colab_type": "text"
      },
      "source": [
        "## French Facebook Muse Embeddings\n",
        "download from facebbok and copy to drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bLbFfNewaHf7",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "f7ef380b-bd79-4b51-978c-080fd0d68a34"
      },
      "source": [
        "%%time\n",
        "\n",
        "french = True\n",
        "\n",
        "if french:\n",
        "\n",
        "    # subset of MUSE Embedding\n",
        "    fr_muse = pickle.load( open(path + '/embeddings/fr_slim_embed.p', \"rb\" ) )\n",
        "    # subset of MUSE Embedding extendet by oov \n",
        "    fr_muse_ext = pickle.load( open(path + '/embeddings/fr_slim_embed_ext.p', \"rb\" ) )\n",
        "    # Embedding trained on supermarket data\n",
        "    fr_own_ft = pickle.load( open(path + '/embeddings/fr_embed_trained.p', \"rb\" ) )\n",
        "    #original MUSE Embedding (takes 5-10 min to load)\n",
        "    #fr_model = KeyedVectors.load_word2vec_format('/content/gdrive/My Drive/Thesis_ecb_ecoicop/embeddings/wiki.fr.vec')\n"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 447 ms, sys: 111 ms, total: 558 ms\n",
            "Wall time: 3.96 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dOgJ8t4brzG0",
        "colab_type": "text"
      },
      "source": [
        "## German Facebook Muse Embeddings\n",
        "download from facebook and copy to drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3L6oHWV7aMGD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "8cf4ab3a-bc53-4268-aa18-6f2c9c9189a6"
      },
      "source": [
        "%%time\n",
        "\n",
        "german = True\n",
        "if german:\n",
        "    # subset of MUSE Embedding\n",
        "    de_muse = pickle.load( open(path + '/embeddings/de_slim_embed.p', \"rb\" ) )\n",
        "    # subset of MUSE Embedding extendet by oov \n",
        "    de_muse_ext = pickle.load( open(path + '/embeddings/de_slim_embed_ext.p', \"rb\" ) )\n",
        "    # Embedding trained on supermarket data\n",
        "    de_own_ft = pickle.load( open(path + '/embeddings/de_embed_trained.p', \"rb\" ) )\n",
        "    #original MUSE Embedding (takes 5-10 min to load)\n",
        "    #de_model = KeyedVectors.load_word2vec_format('/content/gdrive/My Drive/Thesis_ecb_ecoicop/embeddings/wiki.de.vec')"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 762 ms, sys: 228 ms, total: 990 ms\n",
            "Wall time: 6.12 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Afrk_bn4tser",
        "colab_type": "text"
      },
      "source": [
        "# Test Similarity of Embeddings"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AZiDxP1LTSHZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import numpy as np\n",
        "\n",
        "def cosine_similarity( vec_a, vec_b):\n",
        "    \"\"\"Compute cosine similarity between vec_a and vec_b\"\"\"\n",
        "    return np.dot(vec_a, vec_b) / \\\n",
        "        (np.linalg.norm(vec_a) * np.linalg.norm(vec_b))\n",
        "\n",
        "def repla(w):\n",
        "    \"\"\" multi-replace by dict \"\"\"\n",
        "    for a,b in rep_dict.items():\n",
        "        w = w.replace(a,b)\n",
        "    return w\n",
        "  \n",
        "rep_dict = {'.':' ',\n",
        "                ',': ' ',\n",
        "                '&': ' ',\n",
        "                '-': ' ',\n",
        "                '/': ' ',\n",
        "                'ü': 'ue',\n",
        "                'ä': 'ae',\n",
        "                'ö': 'oe',\n",
        "                'ß': 'ss',\n",
        "                'ê': 'e',\n",
        "                'é': 'e',\n",
        "                'è': 'e',\n",
        "                'â': 'a',\n",
        "                'á': 'a',\n",
        "                'à': 'a',\n",
        "                'œ': 'ae',\n",
        "                'î': 'i',\n",
        "                'ç':'c',\n",
        "                'ô':'o',\n",
        "                '%': ' percent ',\n",
        "                '1': ' one ',\n",
        "                '2': ' two ',\n",
        "                '3': ' three ',\n",
        "                '4': ' four ',\n",
        "                '5': ' five ',\n",
        "                '6': ' six ',\n",
        "                '7': ' seven ',\n",
        "                '8': ' eigth ',\n",
        "                '9': ' nine ',\n",
        "                '0': ' zero ',\n",
        "                ' l ':' liter ',\n",
        "                ' ml ':' liter '\n",
        "                }"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8B6O8oe0S7zq",
        "colab_type": "text"
      },
      "source": [
        "### test dict"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ff8lj0qe4jUp",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "762a40cc-6999-40ec-df6e-5db23179c4ed"
      },
      "source": [
        "de_fr_dict = {\n",
        "    'lebensmittel':'aliments',\n",
        "    'getränke':'boissons',\n",
        "    'obst':'fruit',\n",
        "    'gemüse':'légume',\n",
        "    'reis':'riz',\n",
        "    'nuss':'noix',\n",
        "    #'kartoffel':'pomme de terre',\n",
        "    'apfel':'pomme',\n",
        "    'erdbeere':'fraise',\n",
        "    'banane':'banane',\n",
        "    'salat':'salade',\n",
        "    'garnelen':'crevettes',\n",
        "    'muscheln':'coquilles',\n",
        "    'getreide':'céréales',\n",
        "    'mehl':'mer',\n",
        "    'wurzel':'racine',\n",
        "    'fleisch':'viande',\n",
        "    'innereien':'abats',\n",
        "    #'süßigkeiten':'friandise',\n",
        "    #'quark':'quark',\n",
        "    'salz':'sel',\n",
        "    'pfeffer':'poivre',\n",
        "    'erdnuss':'arachide',\n",
        "    'frühstück':'déjeuner',\n",
        "    #'abendessen':'dîner',\n",
        "    'nudeln':'nouilles',\n",
        "    'soße':'sauce',\n",
        "    'zucker':'sucre',\n",
        "    'schokolade':'chocolat',\n",
        "    'kakao':'cacao',\n",
        "    'fett':'graisse',\n",
        "    'pizza':'pizza',\n",
        "    'gewürze':'épices',\n",
        "    'eis':'glace',\n",
        "    'eier':'œufs',\n",
        "    'baby':'bébé',\n",
        "    #'alkohol':' alcool',\n",
        "    'honig':'miel',\n",
        "    'senf':'moutarde',\n",
        "    'tomate':'tomate',\n",
        "    'saft':'jus',\n",
        "    'oliven':'olive',\n",
        "    'pilze':'champignons',\n",
        "    'kaffee': 'café',\n",
        "    'tee': 'thé',\n",
        "    'wurst' : 'saucisse',\n",
        "    #'konfitüre': 'confiture',\n",
        "    'käse': 'fromage',\n",
        "    'gebäck':  'biscuits',\n",
        "    'eis' : 'glace',\n",
        "    'milch': 'lait',\n",
        "    'brot':  'pain',\n",
        "    'suppe': 'soupe',\n",
        "    'huhn':  'poulet',\n",
        "    'fisch':  'poisson',\n",
        "    'butter': 'beurre',\n",
        "    #'kuchen': 'gâteau',\n",
        "    'schinken':  'jambon',\n",
        "    'saft': 'jus',\n",
        "    'fleisch': 'viande',\n",
        "    'zucker': 'sucre',\n",
        "    'schwein': 'porc',\n",
        "    #   'rind': 'bétail',\n",
        "    'pizza': 'pizza',\n",
        "    #   'öl': 'huile', \n",
        "    'eier': 'oeufs',\n",
        "    'wasser': 'eau',\n",
        "    'bier': 'bière',\n",
        "    'getrocknet':'séché',\n",
        "    'gefroren':'gelé',\n",
        "    'frisch':'frais',\n",
        "    'zero':'zero',\n",
        "    'one':'one',\n",
        "    'two':'two',\n",
        "    'three':'three',\n",
        "    'four':'four',\n",
        "    'five':'five',\n",
        "    'six':'six',\n",
        "    'seven':'seven',\n",
        "    'eight':'eight',\n",
        "    'nine':'nine',\n",
        "    'verpackung':'emballage',\n",
        "    'sorten':'variétés',\n",
        "    'alkoholische':'alcoolique',\n",
        "    'milcherzeugnisse':'laitiers',\n",
        "    'pfand':'dépôt',\n",
        "    'zutaten':'ingrédients',\n",
        "    'flasche':'bouteille',\n",
        "    'kauf':'achat',\n",
        "    'weine':'vins',\n",
        "    'fruechte':'fruits',\n",
        "    'spirituosen':'spiritueux',\n",
        "    'aufstriche':'tartiner',\n",
        "    #'tiefkuehlkost':'surgelés',\n",
        "    'fertiggerichte':'préparés',\n",
        "    'bio':'bio',\n",
        "    'liter':'liter',\n",
        "    'hygiene':'hygiene',\n",
        "    'haus':'maison',\n",
        "    'textilien':'textile',\n",
        "    'schön':'beaute'\n",
        "}\n",
        "len(de_fr_dict)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "85"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iVPsdXcJt317",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 214
        },
        "outputId": "157b8b86-36cf-4528-ae3e-74587bb4e255"
      },
      "source": [
        "similar = []\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(fr_muse[fr],de_muse[de]))\n",
        "np.mean(similar)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-4-4c343b33dd2d>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      3\u001b[0m     \u001b[0mfr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m     \u001b[0mde\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mde\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m     \u001b[0msimilar\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcosine_similarity\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfr_muse\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mfr\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mde_muse\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mde\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msimilar\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'fr_muse' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3yJdm2SdTAE4",
        "colab_type": "text"
      },
      "source": [
        "### fucntions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DjXwV9TFoDvO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def similar(word,embed,no=5):\n",
        "    mean =  np.array(embed[word])\n",
        "    limited = np.array(list(embed.values()))\n",
        "    dists = np.dot(limited, mean)\n",
        "    w2d = dict(zip(list(embed.keys()),list(dists)))\n",
        "    dist_milch = [(k,v) for k, v in sorted(w2d.items(), key=lambda item: item[1], reverse=True)]\n",
        "    return dist_milch[1:no+1]\n",
        "\n",
        "def similar_v(v,embed,no=5):\n",
        "    limited = np.array(list(embed.values()))\n",
        "    dists = np.dot(limited, v)\n",
        "    w2d = dict(zip(list(embed.keys()),list(dists)))\n",
        "    dist_milch = [(k,v) for k, v in sorted(w2d.items(), key=lambda item: item[1], reverse=True)]\n",
        "    return dist_milch[1:no+1]\n",
        "  \n",
        "def shared_dict(embed_src,embed_tgt):\n",
        "    in_both_words = [w for w in embed_src if w in embed_tgt]\n",
        "    im_both_dict = {}\n",
        "    for w in in_both_words:\n",
        "        im_both_dict[w] = embed_src[w]\n",
        "    return im_both_dict\n",
        "\n",
        "def similar_both(word,embed_src,embed_tgt,im_both_dict,no=5):\n",
        "    \n",
        "    mean =  np.array(embed_src[word])\n",
        "    limited = np.array(list(im_both_dict.values()))\n",
        "    dists = np.dot(limited, mean)\n",
        "    w2d = dict(zip(list(im_both_dict.keys()),list(dists)))\n",
        "    dist_milch = [[k,v] for k, v in sorted(w2d.items(), key=lambda item: item[1], reverse=True)]\n",
        "\n",
        "    new_vec = np.zeros(mean.shape[0])\n",
        "    v_sum = 0\n",
        "    for k,v in dist_milch[:no]:\n",
        "        new_vec += embed_tgt[k] * v\n",
        "        v_sum += v\n",
        "    new_vec = new_vec / v_sum\n",
        "    new_vec /= np.linalg.norm(new_vec)\n",
        "    return new_vec"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "THY4vm3fcl0q",
        "colab_type": "text"
      },
      "source": [
        "# Train own embeddings"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CrCdt3Hacour",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import gensim"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5YFJGF4Ehv4L",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "04a653d2-1299-4109-df28-da92079b9097"
      },
      "source": [
        "!wget 'https://raw.githubusercontent.com/ELehmann91/Thesis_Multilingual_Transferlearning/master/model_helper.py'\n",
        "import model_helper"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-06-26 19:05:25--  https://raw.githubusercontent.com/ELehmann91/Thesis_Multilingual_Transferlearning/master/model_helper.py\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.0.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 11797 (12K) [text/plain]\n",
            "Saving to: ‘model_helper.py’\n",
            "\n",
            "\rmodel_helper.py       0%[                    ]       0  --.-KB/s               \rmodel_helper.py     100%[===================>]  11.52K  --.-KB/s    in 0s      \n",
            "\n",
            "2020-06-26 19:05:25 (117 MB/s) - ‘model_helper.py’ saved [11797/11797]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0_jgSYGamO0k",
        "colab_type": "text"
      },
      "source": [
        "## German"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yKLHSXiGhL0y",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "3d748b58-dd9c-4638-fea9-b3d6edf06792"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "data_path = '/data/'#\n",
        "file_path = 'normalized/norm_de.csv'\n",
        "\n",
        "if True:\n",
        "    # german data\n",
        "    df_de = pd.read_csv(path+data_path+file_path,sep='|',index_col=False)\n",
        "    df_at = pd.read_csv(path+data_path+'normalized/norm_at.csv',sep='|',index_col=False)\n",
        "    df_de.append(df_at)\n",
        "    len(df_de)\n",
        "\n",
        "    de_voc = model_helper.vocab(df_in=df_de)\n",
        "    de_documents = de_voc.get_list(token=True)\n",
        "    de_voc_count = de_voc.get_vocab(count=True)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py:2718: DtypeWarning: Columns (5,8,17,18,98) have mixed types.Specify dtype option on import or set low_memory=False.\n",
            "  interactivity=interactivity, compiler=compiler, result=result)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tDNLkmX4BUP9",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "f18a3525-060d-4375-91a1-7f72f6ad75e9"
      },
      "source": [
        "len(df_de)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "505395"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fb3WjeYTJ3OH",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "6259fc5e-cffd-4cc1-faca-e0f0b415cd59"
      },
      "source": [
        "i=0\n",
        "for word in de_voc_count.keys():\n",
        "    if word in de_muse_ext.keys():\n",
        "        i+=1\n",
        "print(i,len(de_voc_count),i/len(de_voc_count))"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "60667 114056 0.531905379813425\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7TdadNIZ8KEh",
        "colab_type": "text"
      },
      "source": [
        "### W2V"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zL2kWt0GhMHV",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "76b8ef71-6336-4339-f5a5-5e78aae4529d"
      },
      "source": [
        "%%time\n",
        "if False:\n",
        "    w2v_de = gensim.models.Word2Vec (de_documents, size=150, window=10, min_count=2, workers=10)\n",
        "    w2v_de.train(de_documents,total_examples=len(de_documents),epochs=10)\n",
        "    w1 = \"milch\"\n",
        "    print(w2v_de.wv.most_similar (positive=w1))\n",
        "\n",
        "    de_own_ft_w2v = {} \n",
        "    for word in [k for k, v in sorted(de_voc_count.items(), key=lambda item: item[1],reverse=True)]:\n",
        "        try: de_own_ft_w2v[word] = w2v_de.wv[word]\n",
        "        except:\n",
        "          print(word)\n",
        "    x_tgt = list(de_own_ft_w2v.values())\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "    de_own_ft = dict(zip(list(de_own_ft.keys()),x_tgt))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 4 µs, sys: 0 ns, total: 4 µs\n",
            "Wall time: 7.15 µs\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SqGn6lt88N41",
        "colab_type": "text"
      },
      "source": [
        "### Fasttext"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OhMCGaXW_KAl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from gensim.corpora.wikicorpus import WikiCorpus\n",
        "from gensim.corpora import MmCorpus\n",
        "import gensim\n",
        "import multiprocessing\n",
        "from gensim.models.word2vec import Word2Vec"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FoxU0e9E3zeV",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "c50f5f79-c1da-4a91-8418-d29ddfcad093"
      },
      "source": [
        "if False:\n",
        "#!wget https://dumps.wikimedia.org/dewiki/latest/dewiki-latest-pages-articles1.xml-p1p262468.bz2 #dewiki-latest-pages-articles.xml.bz2\n",
        "    %time wiki = WikiCorpus('dewiki-latest-pages-articles1.xml-p1p262468.bz2')\n",
        "\n",
        "    class MySentences(object): \n",
        "        def __iter__(self): \n",
        "            for text in wiki.get_texts(): \n",
        "                #yield [word.decode() for word in text] \n",
        "                yield [word.lower() for word in text] \n",
        "    sentences = MySentences()\n",
        "\n",
        "    import multiprocessing\n",
        "    from gensim.models import FastText\n",
        "    sentences = MySentences()\n",
        "    params = {'size': 300, 'window': 6, 'iter':5,'word_ngrams':0,'sg':1,'min_count':10,'sorted_vocab':1, \n",
        "              'workers': max(1, multiprocessing.cpu_count()), 'sample': 1e-3,}\n",
        "\n",
        "    ft_de = FastText(**params)\n",
        "    ft_de.build_vocab(sentences)\n",
        "    ft_de.train(sentences, total_examples=ft_de.corpus_count, epochs=20)\n",
        "\n",
        "    #ft_de.save('wiki.ft_300.model')\n",
        "    ft_de.build_vocab(de_documents, update=True)\n",
        "    ft_de.train(de_documents,len(de_documents),epochs=40)\n",
        "\n",
        "    ft_de.most_similar('banane')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-06-22 18:14:43--  https://dumps.wikimedia.org/dewiki/latest/dewiki-latest-pages-articles1.xml-p1p262468.bz2\n",
            "Resolving dumps.wikimedia.org (dumps.wikimedia.org)... 208.80.154.7, 2620:0:861:1:208:80:154:7\n",
            "Connecting to dumps.wikimedia.org (dumps.wikimedia.org)|208.80.154.7|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 473042912 (451M) [application/octet-stream]\n",
            "Saving to: ‘dewiki-latest-pages-articles1.xml-p1p262468.bz2’\n",
            "\n",
            "dewiki-latest-pages 100%[===================>] 451.13M  4.55MB/s    in 1m 43s  \n",
            "\n",
            "2020-06-22 18:16:26 (4.36 MB/s) - ‘dewiki-latest-pages-articles1.xml-p1p262468.bz2’ saved [473042912/473042912]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TqGH36bA8P1u",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 105
        },
        "outputId": "defa613d-123f-424b-d848-b54e3f66c3bd"
      },
      "source": [
        "%%time\n",
        "import multiprocessing\n",
        "from gensim.models import FastText\n",
        "import pickle\n",
        "\n",
        "if True:\n",
        "    params = {'size': 300, 'window': 6, 'iter':20,'word_ngrams':1,'sg':1,'min_count':20,'sorted_vocab':1, \n",
        "              'workers': max(1, multiprocessing.cpu_count()), 'sample': 1e-3,}\n",
        "\n",
        "    ft_de = FastText(**params)\n",
        "    ft_de.build_vocab(de_documents)\n",
        "    ft_de.train(de_documents, total_examples=ft_de.corpus_count, epochs=20)\n",
        "\n",
        "    ft_de.wv.most_similar('milch')\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 2h 51min 10s, sys: 13.1 s, total: 2h 51min 24s\n",
            "Wall time: 43min 45s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8ZI-FQfFpVP-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "voc_sub = []\n",
        "for word in ft_de.vocab:\n",
        "    for i in range (len(word)-2):\n",
        "        voc_sub.append(word[i:i+3])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4PwGOT4fpJTz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "sub_vocab = [b for b in ft_de.vocab]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BsaRWHM-pJcM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "    de_own_ft = {} \n",
        "    for word in [k for k, v in sorted(de_voc_count.items(), key=lambda item: item[1],reverse=True)]:\n",
        "        try: de_own_ft[word] = ft_de.wv[word]\n",
        "        except:\n",
        "          print(word)\n",
        "\n",
        "    x_tgt = list(de_own_ft.values())\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "    de_own_ft = dict(zip(list(de_own_ft.keys()),x_tgt))\n",
        "    pickle.dump(de_own_ft, open(path + '/embeddings/de_embed_trained.p', \"wb\" ))\n",
        "\n",
        "else:\n",
        "    de_own_ft = pickle.load( open(path + '/embeddings/de_embed_trained.p', \"rb\" ))\n",
        "    x_tgt = list(de_own_ft.values())\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "    de_own_ft = dict(zip(list(de_own_ft.keys()),x_tgt))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Px-zXuhuyoYE",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "894b732d-70b3-4c2c-b100-63ea9c201b2d"
      },
      "source": [
        "cosine_similarity(de_own_ft['milch'],de_own_ft['joghurt'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.37362045"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LjZoE3ZtHD-E",
        "colab_type": "text"
      },
      "source": [
        "### Translation of most freq words"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YE6RHMDGDinA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import re\n",
        "def prepro(line,rep_dict):\n",
        "    if isinstance(line,str):\n",
        "        text_str = ' '.join(str(t) for t in line.split())\n",
        "        text_str = text_str.lower()\n",
        "        for a,b in rep_dict.items():\n",
        "            text_str = text_str.replace(a,b)\n",
        "        text_str = re.sub('[^a-zäöüàáâéèêßœ]+', ' ', text_str)\n",
        "    else: \n",
        "        text_str = str(line)\n",
        "        print(line)\n",
        "    return text_str\n",
        "\n",
        "class vocab(object):\n",
        "    '''\n",
        "    takes the standardized dataframe and gives out vocab, with index, with word counts\n",
        "    the text in rows as list of string or list of token and\n",
        "    builds a subset of embedding including out ov vocabulary items\n",
        "    '''\n",
        "    def __init__(self, df_in = None):\n",
        "        self.df_in = df_in\n",
        "        self.lang = str(self.df_in['lang'].iloc[1])\n",
        "        self.df_in['text'] =  self.df_in['name'].fillna('unknown') + ' ' + \\\n",
        "                              self.df_in['categ'].fillna('unknown') + ' ' + \\\n",
        "                              self.df_in['words_from_url'].fillna('unknown') \n",
        "    \n",
        "    def get_list(self,preprocess=True,token=False,one_obj =False):\n",
        "        if preprocess:\n",
        "            list_str = [prepro(line,rep_dict) for line in self.df_in['text']]\n",
        "        else:\n",
        "            list_str = [line for line in self.df_in['text']]\n",
        "        if token:\n",
        "            list_str = [[word for word in line.split()] for line in list_str]\n",
        "        if one_obj:\n",
        "            temp_list = []\n",
        "            for line in list_str:\n",
        "                tokens = [word for word in line.split()]\n",
        "                temp_list.extend(tokens)\n",
        "            list_str = temp_list\n",
        "        return list_str\n",
        "\n",
        "    def get_vocab(self,index=False,count=False):\n",
        "        vocab = {}\n",
        "        i=1\n",
        "        for tok in self.get_list(one_obj=True):\n",
        "            if tok in vocab and count:\n",
        "                vocab[tok] += 1\n",
        "            else:\n",
        "                if count:\n",
        "                    vocab[tok] = 1\n",
        "                else:\n",
        "                    vocab[tok] = i\n",
        "                    i += 1\n",
        "        if index or count:\n",
        "            return {k: v for k, v in sorted(vocab.items(), key=lambda item: item[1],reverse=True)}\n",
        "        else: \n",
        "            return list(vocab.keys())"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1RsqIr8MDKN5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 156
        },
        "outputId": "e8fef80c-3276-44d6-9346-839494ea6041"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "data_path = '/data/'#\n",
        "#file_path = 'fra/carrfour_trans_pred.csv'\n",
        "file_path = 'normalized/norm_fr.csv'\n",
        "file_path2 = 'normalized/norm_de.csv'\n",
        "file_path3 = 'at/norm_at.csv'\n",
        "#file_path = 'edeka_pred.csv'\n",
        "\n",
        "# french data\n",
        "df_fr = pd.read_csv(path+data_path+file_path,sep='|',index_col=False)\n",
        "\n",
        "# only already labeled\n",
        "df_fr = df_fr[df_fr['cc5'].isna()==False]\n",
        "df_fr = df_fr[df_fr['shop'].isin( ['carrefour','auchan'])]#,'banque_de_france'])]\n",
        "df_fr['cc5'] = df_fr['cc5'].apply(lambda x: '9999_Non-Food' if int(str(x)[0])>2 else x)\n",
        "print(len(df_fr))\n",
        "\n",
        "# german data\n",
        "df_de = pd.read_csv(path+data_path+file_path2,sep='|',index_col=False)\n",
        "# only already labeled\n",
        "df_de = df_de[df_de['cc5'].isna()==False]\n",
        "print(len(df_de))\n",
        "# austrian data\n",
        "df_at = pd.read_csv(path+data_path+file_path3,sep='|',index_col=False)\n",
        "# only already labeled\n",
        "df_at = df_at[df_at['cc5'].isna()==False]\n",
        "df_de = df_de.append(df_at)\n",
        "print(len(df_de))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py:2718: DtypeWarning: Columns (0,3,4,5,6,7,8,9,10,11,12,13,14,16,17,18,19) have mixed types.Specify dtype option on import or set low_memory=False.\n",
            "  interactivity=interactivity, compiler=compiler, result=result)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "7923\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py:2718: DtypeWarning: Columns (5,8,17,18,98) have mixed types.Specify dtype option on import or set low_memory=False.\n",
            "  interactivity=interactivity, compiler=compiler, result=result)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "21903\n",
            "23199\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WaIms7jZEZGJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "de_voc = vocab(df_in=df_de)\n",
        "#de_documents = de_voc.get_list(token=True)\n",
        "de_voc_count = de_voc.get_vocab(count=True)\n",
        "\n",
        "fr_voc = vocab(df_in=df_fr)\n",
        "#de_documents = de_voc.get_list(token=True)\n",
        "fr_voc_count = de_voc.get_vocab(count=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "svK78lLwFm9X",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "b438c691-e6b4-4e2f-a466-9da2a0a42b0b"
      },
      "source": [
        "len(fr_voc_count)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "13608"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-BnwwOrOHELi",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "960551c0-1aa3-4d7d-8e8f-6c020a11043f"
      },
      "source": [
        "%%capture\n",
        "!pip install --upgrade pip\n",
        "!pip install --upgrade google-api-python-client\n",
        "import getpass\n",
        "APIKEY = getpass.getpass()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "··········\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e79dYOn_NAPt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "## running Translate API\n",
        "from googleapiclient.discovery import build\n",
        "service = build('translate', 'v2', developerKey=APIKEY)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jtq-dnluIScG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pickle\n",
        "trans_list  = pickle.load( open(path + '/data/de_2000_fr.p', \"rb\" ) ) #\n",
        "#pickle.dump(trans_list, open(path + '/data/de_2000_fr.p', \"wb\" ))\n",
        "\n",
        "def repla(w):\n",
        "    \"\"\" multi-replace by dict \"\"\"\n",
        "    for a,b in rep_dict.items():\n",
        "        w = w.replace(a,b).lower()\n",
        "    return w\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4Rjnr_OeIevD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "a0a7f80a-e91b-40d2-afa1-929d4d61774d"
      },
      "source": [
        "for w_de in list(de_voc_count.keys())[:5000]:\n",
        "    if w_de in trans_list.keys():\n",
        "        trans_list[w_de] = trans_list[w_de]\n",
        "    else:\n",
        "        w_fr = service.translations().list(source='de', target='fr', q=w_de).execute()\n",
        "        trans_list[w_de] = w_fr['translations'][0]['translatedText']\n",
        "\n",
        "len(trans_list)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "5737"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 59
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "E1oho8U3Gt-q",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "2690190d-f292-4a67-c7c6-d88227fe1fba"
      },
      "source": [
        "trans_list_rev = {b:a for a,b in trans_list.items()}\n",
        "\n",
        "for w_fr in list(fr_voc_count.keys())[:5000]:\n",
        "    if w_fr not in trans_list_rev.keys():\n",
        "        w_de = service.translations().list(source='fr', target='de', q=w_fr).execute()\n",
        "        w_de = w_de['translations'][0]['translatedText']\n",
        "        if w_de not in trans_list.keys():\n",
        "            trans_list[w_de] = w_fr\n",
        "\n",
        "len(trans_list)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "7200"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 62
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qz6ikrs6NcsM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pickle\n",
        "\n",
        "pickle.dump(trans_list, open(path + '/data/de_5000_fr.p', \"wb\" ))\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "STHPa8lQJy7T",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "fr_muse_ext = pickle.load( open(path + '/embeddings/fr_slim_embed_ext.p', \"rb\" ) )\n",
        "de_muse_ext = pickle.load( open(path + '/embeddings/de_slim_embed_ext.p', \"rb\" ) )"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KT8wZ1-7NZU7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for de_w,fr_w in trans_list.items():\n",
        "    fr_w = repla(fr_w)\n",
        "    de_w = repla(de_w)\n",
        "\n",
        "    if de_w in de_muse_ext:\n",
        "        de_emb = de_muse_ext[de_w]\n",
        "    elif de_w in fr_muse_ext:\n",
        "        de_emb = fr_muse_ext[de_w]\n",
        "    else:\n",
        "        de_emb = np.zeros(300)\n",
        "\n",
        "    if len(fr_w.split(' '))>1:\n",
        "            i=0\n",
        "            emb_fr = np.zeros(300)\n",
        "            #print(fr_w.split(' '))\n",
        "            for w in fr_w.split(' '):\n",
        "                w=repla(w)\n",
        "                if len(w)>2: \n",
        "                    try:\n",
        "                        emb = fr_git_embed[w] \n",
        "                        i +=1\n",
        "                        #print(w)\n",
        "                        emb_fr += emb\n",
        "                    except:\n",
        "                        next\n",
        "            if i>0:\n",
        "                emb_fr = emb_fr/i\n",
        "                #print(emb_fr.shape)\n",
        "                fr_emb = emb_fr\n",
        "    else:\n",
        "        if fr_w in fr_muse_ext:\n",
        "            fr_emb = fr_muse_ext[fr_w]\n",
        "        #elif fr_w in de_muse_ext:\n",
        "        #    fr_emb = de_muse_ext[fr_w]\n",
        "        else:\n",
        "            fr_emb = np.zeros(300)\n",
        "\n",
        "    fr_muse_ext[de_w] = fr_emb\n",
        "    de_muse_ext[fr_w] = de_emb\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N_abRR6tNReI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "pickle.dump(fr_muse_ext, open(path + '/embeddings/fr_slim_embed_ext_voc_share.p', \"wb\" ))\n",
        "pickle.dump(de_muse_ext, open(path + '/embeddings/de_slim_embed_ext_voc_share.p', \"wb\" ))\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6Bf-uaBumRZ0",
        "colab_type": "text"
      },
      "source": [
        "## French"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rvzvk7dqmAf9",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        },
        "outputId": "e76cc13b-7760-4d61-f759-d5fcdd5a2cbc"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "data_path = '/data/'#\n",
        "file_path = 'normalized/norm_fr.csv'\n",
        "\n",
        "if True:\n",
        "# french data\n",
        "    df_fr = pd.read_csv(path+data_path+file_path,sep='|',index_col=False)\n",
        "    fr_voc = model_helper.vocab(df_in=df_fr)\n",
        "    documents_fr = fr_voc.get_list(token=True)\n",
        "    fr_voc_count = fr_voc.get_vocab(count=True)\n",
        "    len(documents_fr)"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py:2718: DtypeWarning: Columns (0,3,4,5,6,7,8,9,10,11,12,13,14,16,17,18,19) have mixed types.Specify dtype option on import or set low_memory=False.\n",
            "  interactivity=interactivity, compiler=compiler, result=result)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k9V6Uw6SBxAj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "cac4ab13-7b03-4176-fed3-05fc63549923"
      },
      "source": [
        "i=0\n",
        "for word in fr_voc_count.keys():\n",
        "    if word in fr_muse.keys():\n",
        "        i+=1\n",
        "print(i,len(fr_voc_count),i/len(fr_voc_count),len(df_fr))"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "41211 55047 0.7486511526513706 329149\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "J-HL27XLAVFo",
        "colab_type": "text"
      },
      "source": [
        "### W2V"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VAMCCs4BmYzy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "25bcc22f-aec7-4db2-91d4-e25fd3d8a788"
      },
      "source": [
        "%%time\n",
        "if False:\n",
        "    w2v_fr = gensim.models.Word2Vec (documents_fr, size=150, window=10, min_count=2, workers=10)\n",
        "    w2v_fr.train(documents_fr,total_examples=len(documents_fr),epochs=10)\n",
        "\n",
        "    print(w2v_fr.wv.most_similar('thé'))\n",
        "\n",
        "    fr_emb_ord_w2v = {} \n",
        "    for word in [k for k, v in sorted(fr_voc_count.items(), key=lambda item: item[1],reverse=True)]:\n",
        "        try:\n",
        "            fr_emb_ord_w2v[word] = w2v_fr.wv[word]\n",
        "        except:\n",
        "            print(word)\n",
        "    x_tgt = list(de_emb_ord_w2v.values())\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "    de_emb_ord_w2v = dict(zip(list(de_emb_ord_w2v.keys()),x_tgt))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 4 µs, sys: 0 ns, total: 4 µs\n",
            "Wall time: 6.68 µs\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vRucIuu3BwSy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6DIoeYYpAYl7",
        "colab_type": "text"
      },
      "source": [
        "### FastText"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lx0RsMIPAX2y",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        },
        "outputId": "2562735a-2f98-4294-baa6-7277d5a07fc1"
      },
      "source": [
        "%%time\n",
        "from gensim.models import FastText\n",
        "import multiprocessing\n",
        "import pickle\n",
        "\n",
        "if False:\n",
        "    params = {'size': 300, 'window': 8, 'iter':20,'word_ngrams':0,'sg':1,'min_count':10,'sorted_vocab':1 ,\n",
        "              'workers': max(1, multiprocessing.cpu_count() - 1\n",
        "                            ), 'sample': 1e-3,}\n",
        "\n",
        "    ft_fr = FastText(**params)\n",
        "    ft_fr.build_vocab(documents_fr)\n",
        "    ft_fr.train(documents_fr, total_examples=ft_fr.corpus_count, epochs=20)\n",
        "\n",
        "    print(ft_fr.wv.most_similar('lait'))\n",
        "\n",
        "    fr_own_ft = {} \n",
        "    for word in [k for k, v in sorted(fr_voc_count.items(), key=lambda item: item[1],reverse=True)]:\n",
        "        try:\n",
        "            fr_emb_ord[word] = ft_fr.wv[word]\n",
        "        except:\n",
        "            print(word)\n",
        "\n",
        "    x_tgt = list(fr_own_ft.values())\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "    fr_own_ft = dict(zip(list(fr_own_ft.keys()),x_tgt))\n",
        "    pickle.dump( fr_own_ft, open(path + '/embeddings/fr_embed_trained.p', \"wb\" ) )\n",
        "else:\n",
        "    fr_own_ft = pickle.load( open(path + '/embeddings/fr_embed_trained.p', \"rb\" ) )\n",
        "    x_tgt = list(fr_own_ft.values())\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "    fr_own_ft = dict(zip(list(fr_own_ft.keys()),x_tgt))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 360 ms, sys: 86.3 ms, total: 446 ms\n",
            "Wall time: 452 ms\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PEQ7b6JlmFip",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "542517a4-12f7-4b91-b16f-18b8b58d1e06"
      },
      "source": [
        "cosine_similarity(fr_emb_ord['lait'],fr_emb_ord['fromage'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.219518"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LOxV6JJUY10u",
        "colab_type": "text"
      },
      "source": [
        "# Align Embeddings"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tVoKT9ln_2FG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "%%capture\n",
        "!pip install POT"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RgkQh-FRHKHH",
        "colab_type": "text"
      },
      "source": [
        "## Extend Vocab of pretrained Muse by own trained using KNN"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yfYpXwlpHN-g",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def similar(word,embed,no=5):\n",
        "    ''' get the n closed neighbours for a word '''\n",
        "    mean =  np.array(embed[word])\n",
        "    limited = np.array(list(embed.values()))\n",
        "    dists = np.dot(limited, mean)\n",
        "    w2d = dict(zip(list(embed.keys()),list(dists)))\n",
        "    dist_milch = [(k,v) for k, v in sorted(w2d.items(), key=lambda item: item[1], reverse=True)]\n",
        "    return dist_milch[1:no+1]\n",
        "\n",
        "def similar_v(v,embed,no=5):\n",
        "    ''' get the n closed neighbours for a vector '''\n",
        "    limited = np.array(list(embed.values()))\n",
        "    dists = np.dot(limited, v)\n",
        "    w2d = dict(zip(list(embed.keys()),list(dists)))\n",
        "    dist_milch = [(k,v) for k, v in sorted(w2d.items(), key=lambda item: item[1], reverse=True)]\n",
        "    return dist_milch[1:no+1]\n",
        "  \n",
        "def shared_dict(embed_src,embed_tgt):\n",
        "    ''' words are in A and B dict'''\n",
        "    in_both_words = [w for w in embed_src if w in embed_tgt]\n",
        "    im_both_dict = {}\n",
        "    for w in in_both_words:\n",
        "        im_both_dict[w] = embed_src[w]\n",
        "    return im_both_dict\n",
        "\n",
        "def similar_both(word,embed_src,embed_tgt,im_both_dict,no=5):\n",
        "    ''' transfer vocab from one space to another by its neighbours '''\n",
        "    if word in embed_src:\n",
        "        mean =  np.array(embed_src[word])\n",
        "        limited = np.array(list(im_both_dict.values()))\n",
        "        dists = np.dot(limited, mean)\n",
        "        w2d = dict(zip(list(im_both_dict.keys()),list(dists)))\n",
        "        dist_milch = [[k,v] for k, v in sorted(w2d.items(), key=lambda item: item[1], reverse=True)]\n",
        "\n",
        "        new_vec = np.zeros(mean.shape[0])\n",
        "        v_sum = 0\n",
        "        for k,v in dist_milch[:no]:\n",
        "            new_vec += embed_tgt[k] * v\n",
        "            v_sum += v\n",
        "        new_vec = new_vec / v_sum\n",
        "        new_vec /= np.linalg.norm(new_vec)\n",
        "        return new_vec\n",
        "    else:\n",
        "        return np.zeros(300)\n",
        "\n",
        "def objective(X, Y, R, n=5000):\n",
        "    Xn, Yn = X[:n], Y[:n]\n",
        "    C = -np.dot(np.dot(Xn, R), Yn.T)\n",
        "    P = ot.sinkhorn(np.ones(n), np.ones(n), C, 0.025, stopThr=1e-3)\n",
        "    return 1000 * np.linalg.norm(np.dot(Xn, R) - np.dot(P, Yn)) / n\n",
        "\n",
        "\n",
        "def sqrt_eig(x):\n",
        "    U, s, VT = np.linalg.svd(x, full_matrices=False)\n",
        "    return np.dot(U, np.dot(np.diag(np.sqrt(s)), VT))\n",
        "\n",
        "\n",
        "def align(X, Y, R, lr=10., bsz=200, nepoch=5, niter=1000,\n",
        "          nmax=10000, reg=0.05, verbose=True):\n",
        "    for epoch in range(1, nepoch + 1):\n",
        "        for _it in range(1, niter + 1):\n",
        "            # sample mini-batch\n",
        "            xt = X[np.random.permutation(nmax)[:bsz], :]\n",
        "            yt = Y[np.random.permutation(nmax)[:bsz], :]\n",
        "            # compute OT on minibatch\n",
        "            C = -np.dot(np.dot(xt, R), yt.T)\n",
        "            P = ot.sinkhorn(np.ones(bsz), np.ones(bsz), C, reg, stopThr=1e-3)\n",
        "            # compute gradient\n",
        "            G = - np.dot(xt.T, np.dot(P, yt))\n",
        "            R -= lr / bsz * G\n",
        "            # project on orthogonal matrices\n",
        "            U, s, VT = np.linalg.svd(R)\n",
        "            R = np.dot(U, VT)\n",
        "        bsz *= 2\n",
        "        niter //= 4\n",
        "        if verbose:\n",
        "            print(\"epoch: %d  obj: %.3f\" % (epoch, objective(X, Y, R)))\n",
        "    return R\n",
        "\n",
        "# this function does not work\n",
        "def convex_init(X, Y, niter=100, reg=0.05, apply_sqrt=False):\n",
        "    n, d = X.shape\n",
        "    if apply_sqrt:\n",
        "        X, Y = sqrt_eig(X), sqrt_eig(Y)\n",
        "    K_X, K_Y = np.dot(X, X.T), np.dot(Y, Y.T)\n",
        "    K_Y *= np.linalg.norm(K_X) / np.linalg.norm(K_Y)\n",
        "    K2_X, K2_Y = np.dot(K_X, K_X), np.dot(K_Y, K_Y)\n",
        "    P = np.ones([n, n]) / float(n)\n",
        "    for it in range(1, niter + 1):\n",
        "        G = np.dot(P, K2_X) + np.dot(K2_Y, P) - 2 * np.dot(K_Y, np.dot(P, K_X))\n",
        "        #print(it,G,np.sum(np.isnan(G)))\n",
        "        q = ot.sinkhorn(np.ones(n), np.ones(n), G, reg, stopThr=1e-3)\n",
        "        #print(it,q,np.sum(np.isnan(q)))\n",
        "        alpha = 2.0 / float(2.0 + it)\n",
        "        P = alpha * q + (1.0 - alpha) * P\n",
        "        #print(it,sum(np.isnan(G)),sum(np.isnan(q)),sum(np.isnan(P)))\n",
        "    obj = np.linalg.norm(np.dot(P, K_X) - np.dot(K_Y, P))\n",
        "    #print(sum(np.isnan(X)),sum(np.isnan(Y)),sum(np.isnan(P)))\n",
        "    return procrustes(np.dot(P, X), Y).T\n",
        "\n",
        "def procrustes(X_src, Y_tgt):\n",
        "    U, s, V = np.linalg.svd(np.dot(Y_tgt.T, X_src))\n",
        "    return np.dot(U, V)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "k-kKXLnTHwq2",
        "colab_type": "text"
      },
      "source": [
        "### Out of Vocab words to include to Muse"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "74sSRbcA-TyM",
        "colab_type": "text"
      },
      "source": [
        "How does it work?  \n",
        "* calculate a shared dict\n",
        "* take the n closest words from shared dict\n",
        "* calcluated the weighted average of those "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rlwg6cCPVcMk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_tgt = list(de_own_ft.values())\n",
        "x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "de_own_ft = dict(zip(list(de_own_ft.keys()),x_tgt))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4qN-jzPRVogs",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_tgt = list(fr_own_ft.values())\n",
        "x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "fr_own_ft = dict(zip(list(fr_own_ft.keys()),x_tgt))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yTkupPhsHOP6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 55
        },
        "outputId": "f30da99e-0bd3-47fd-bc21-cc595550f7f1"
      },
      "source": [
        "word = 'suesses'\n",
        "\n",
        "im_both_dict = shared_dict(de_own_ft,de_muse)\n",
        "new_v = similar_both(word,de_own_ft,de_muse,im_both_dict,10)\n",
        "print(word,similar_v(new_v,de_muse))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "suesses [('salzigem', 0.8042585822582351), ('salzigen', 0.7964649717836925), ('salziges', 0.7862433539749346), ('salziger', 0.7816343457101059), ('soesses', 0.7373135381107807)]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VoHb1BDQIGPX",
        "colab_type": "text"
      },
      "source": [
        "### Append to Muse"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pDKprorBIGZZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        },
        "outputId": "3642df43-3541-4913-91b9-55b5aec3965f"
      },
      "source": [
        "%%time\n",
        "from tqdm import tqdm\n",
        "\n",
        "def emb_knn_align(own_ft,muse,voc_count,min_c=25,nn=5):\n",
        "\n",
        "    oov = []\n",
        "    for word,v in [(k,v) for k, v in sorted(voc_count.items(), key=lambda item: item[1],reverse=True) if v > min_c]:\n",
        "        if word not in muse:\n",
        "            oov.append(word)\n",
        "    print('ooV words:',len(oov),oov[1:15])\n",
        "\n",
        "    muse_ext = muse\n",
        "    in_both_dict = shared_dict(own_ft,muse)\n",
        "\n",
        "    for word in tqdm(oov):\n",
        "        if word in own_ft:\n",
        "            new_v = similar_both(word,own_ft,muse,in_both_dict,nn)\n",
        "            muse_ext[word] = new_v\n",
        "    return muse_ext\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 14 µs, sys: 0 ns, total: 14 µs\n",
            "Wall time: 16.9 µs\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lt16yGlAHZLQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 110
        },
        "outputId": "8d7ec8ee-26b6-4328-dddc-2eda3077e969"
      },
      "source": [
        "data_path = '/data/'#\n",
        "file_path = 'normalized/norm_fr.csv'\n",
        "\n",
        "if True:\n",
        "    # french data\n",
        "    df_fr = pd.read_csv(path+data_path+file_path,sep='|',index_col=False)\n",
        "    fr_voc = model_helper.vocab(df_in=df_fr)\n",
        "    fr_voc_count = fr_voc.get_vocab(count=True)\n",
        "\n",
        "    fr_muse_ext = emb_knn_align(fr_own_ft,fr_muse,fr_voc_count,min_c=10,nn=5)\n",
        "    pickle.dump(fr_muse_ext, open(path + '/embeddings/fr_slim_embed_ext.p', \"wb\" ))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py:2718: DtypeWarning: Columns (0,3,4,5,6,7,8,9,10,11,12,13,14,16,17,18,19) have mixed types.Specify dtype option on import or set low_memory=False.\n",
            "  interactivity=interactivity, compiler=compiler, result=result)\n",
            "  0%|          | 1/1708 [00:00<03:55,  7.26it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "ooV words: 1708 ['rafra', 'gumes', 'gerble', 'naturnes', 'chissante', 'multifruits', 'febreze', 'babybio', 'elseve', 'dermatologiquement', 'elastiquee', 'teaux', 'gerlinea', 'klorane']\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 1708/1708 [07:13<00:00,  3.94it/s]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ygaGX9yYXhvN",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 110
        },
        "outputId": "31fa5065-c58f-4885-b7c4-0bfdea0d2765"
      },
      "source": [
        "data_path = '/data/'#\n",
        "file_path = 'normalized/norm_de.csv'\n",
        "\n",
        "if True:\n",
        "    # german data\n",
        "    df_de = pd.read_csv(path+data_path+file_path,sep='|',index_col=False)\n",
        "    de_voc = model_helper.vocab(df_in=df_de)\n",
        "    de_voc_count = de_voc.get_vocab(count=True)\n",
        "\n",
        "    de_muse_ext = emb_knn_align(de_own_ft,de_muse,de_voc_count,min_c=10,nn=5)\n",
        "    pickle.dump(de_muse_ext, open(path + '/embeddings/de_slim_embed_ext.p', \"wb\" ))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py:2718: DtypeWarning: Columns (5,8,17,18,98) have mixed types.Specify dtype option on import or set low_memory=False.\n",
            "  interactivity=interactivity, compiler=compiler, result=result)\n",
            "  0%|          | 0/8423 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "ooV words: 8423 ['eigth', 'nfuer', 'typename', 'productcontent', 'tiefkuehlkost', 'suessigkeiten', 'wuerzelgemuese', 'wuerste', 'gefluegel', 'meeresfruechte', 'suesses', 'zitrusfruechte', 'kuehlung', 'suesswaren']\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 8423/8423 [44:25<00:00,  3.16it/s]\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CX2eQBEKyNNZ",
        "colab_type": "text"
      },
      "source": [
        "## Single language alignment with dictionary"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sWgES633pa7m",
        "colab_type": "text"
      },
      "source": [
        "Extend Muse embedding by rotation of own trained embedding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zzYeyH0q3avk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# sort embedding\n",
        "def order_dict(dic,voc_count,min_c=20):\n",
        "    dic_ord = {} \n",
        "    for word,v in [(k,v) for k, v in sorted(voc_count.items(), key=lambda item: item[1],reverse=True) if v > min_c]:\n",
        "        word = repla(word)\n",
        "        try:\n",
        "            dic_ord[word] = dic[word]\n",
        "        except:\n",
        "            pass\n",
        "    return dic_ord\n",
        "  \n",
        "def prep_align(tgt_dic,src_dic,translate=False):\n",
        "    # vocab own trained\n",
        "    w_tgt = list(tgt_dic.keys())\n",
        "    # embeddings own trained\n",
        "    x_tgt = np.array(list(tgt_dic.values()))\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "\n",
        "    # vocab own trained\n",
        "    w_src = list(src_dic.keys())\n",
        "    # embeddings own trained\n",
        "    x_src = np.array(list(src_dic.values()))\n",
        "    x_src /= np.linalg.norm(x_src, axis=1)[:, np.newaxis] + 1e-8\n",
        "\n",
        "    # dictionary of words to dictionary of index\n",
        "    if translate:\n",
        "        src2trg = dict(zip([w_src.index(repla(a)) for a in de_fr_dict.keys()],[w_tgt.index(repla(b)) for b in de_fr_dict.values()]))\n",
        "    else:\n",
        "        src2trg = dict(zip([w_src.index(repla(a)) for a in de_fr_dict.keys()],[w_tgt.index(repla(b)) for b in de_fr_dict.keys()]))\n",
        "    return w_tgt, x_tgt, w_src, x_src, src2trg\n",
        "\n",
        "def run_alignment(x_tgt,x_src,lr,nepoch):\n",
        "    print(\"\\n*** Wasserstein Procrustes ***\\n\")\n",
        "    import math\n",
        "    import time\n",
        "    import ot\n",
        "\n",
        "    print(\"\\nComputing initial mapping with convex relaxation...\")\n",
        "    t0 = time.time()\n",
        "    R0 = convex_init(x_src[:2500], x_tgt[:2500], reg=0.01,niter=20, apply_sqrt=True)\n",
        "    print(\"Done [%03d sec]\" % math.floor(time.time() - t0))\n",
        "\n",
        "    print(\"\\nComputing mapping with Wasserstein Procrustes...\")\n",
        "    t0 = time.time()\n",
        "    R = align(x_src, x_tgt, R0.copy(),lr=lr,  bsz=200, nepoch=nepoch, niter=2500, nmax=10000, reg=0.01, verbose=True)\n",
        "    print(\"Done [%03d sec]\" % math.floor(time.time() - t0))\n",
        "    \n",
        "    x_src_ = x_src / np.linalg.norm(x_src, 2, 1).reshape([-1, 1])\n",
        "\n",
        "    # transform target with learned transformation matrix\n",
        "    x_tgt = np.dot(x_tgt, R.T)\n",
        "    x_tgt = x_tgt / np.linalg.norm(x_tgt, 2, 1).reshape([-1, 1])\n",
        "    return x_tgt,x_src"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3jCU0cm3x5fZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "emb_src = de_muse\n",
        "emb_tgt = de_own_ft"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AJ4Gl0yxx6mJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "similar=[]\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(emb_src[de],emb_tgt[de]))\n",
        "np.mean(similar)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OzfSC37CmkqG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "emb_tgt = order_dict(emb_tgt,de_voc_count)\n",
        "emb_src = order_dict(emb_src,de_voc_count)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IojzzjPfo-fp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "w_tgt, x_tgt, w_src, x_src, src2trg = prep_align(de_own_ft,de_muse,translate=False)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UI81WXreWV9L",
        "colab_type": "text"
      },
      "source": [
        "transform it!"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8NSCH54h0B57",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 333
        },
        "outputId": "227a8952-89ee-4bfe-aaca-cf75eb38a159"
      },
      "source": [
        "x_tgt_trans , x_src_norm = run_alignment(x_tgt,x_src,1500,5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "*** Wasserstein Procrustes ***\n",
            "\n",
            "\n",
            "Computing initial mapping with convex relaxation...\n",
            "Done [046 sec]\n",
            "\n",
            "Computing mapping with Wasserstein Procrustes...\n",
            "epoch: 1  obj: 10.990\n",
            "epoch: 2  obj: 10.793\n",
            "epoch: 3  obj: 10.592\n",
            "epoch: 4  obj: 10.460\n",
            "epoch: 5  obj: 10.415\n",
            "epoch: 6  obj: 10.407\n",
            "epoch: 7  obj: 10.407\n",
            "epoch: 8  obj: 10.407\n",
            "epoch: 9  obj: 10.407\n",
            "epoch: 10  obj: 10.407\n",
            "Done [706 sec]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OQYx-lPZWVJS",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "f11fd226-9658-4c50-9b77-857958dd99d2"
      },
      "source": [
        "similar=[]\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(x_tgt_trans[w_tgt.index(de)],x_src_norm[w_src.index(de)]))\n",
        "    #similar.append(cosine_similarity(de_muse[de],de_own_ft[de]))\n",
        "np.mean(similar)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.205737190285347"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 169
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8zVuYRORWyFo",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 183
        },
        "outputId": "9400569f-e623-445c-a908-86af5adf2769"
      },
      "source": [
        "de = 'kuchen'\n",
        "de_vec = x_tgt_trans[w_tgt.index(de)]\n",
        "similar_v(de_vec,de_own_ft,10)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('torten', 4.3773546),\n",
              " ('teilchen', 3.922698),\n",
              " ('streuselkuchen', 3.8995883),\n",
              " ('kuchenrost', 3.8982143),\n",
              " ('marmorkuchen', 3.8810647),\n",
              " ('kuchenglasur', 3.6762817),\n",
              " ('gugelhupf', 3.5854821),\n",
              " ('plunder', 3.5825305),\n",
              " ('apfelkuchen', 3.5481796),\n",
              " ('blechkuchen', 3.5169294)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 171
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VIllWj48lsHt",
        "colab_type": "text"
      },
      "source": [
        "## Language alignment with dictionary"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "suT3kGQ_l_fE",
        "colab_type": "text"
      },
      "source": [
        "Align Muse embedding by rotation of own trained embedding"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ilboYmNuldN1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import ot"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H4GWau3Oxxmm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "emb_src = de_muse_ext\n",
        "emb_tgt = fr_muse_ext"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9yWL9Nlhl_wL",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "b7e5afca-4240-4413-810e-abe2a3807460"
      },
      "source": [
        "similar=[]\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(emb_src[de],emb_tgt[fr]))\n",
        "np.mean(similar)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.5056989084314625"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZTd6_iTpl_t5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "emb_tgt = order_dict(emb_tgt,fr_voc_count,1)\n",
        "emb_src = order_dict(emb_src,de_voc_count)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NAq88RU3l_rn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "w_tgt, x_tgt, w_src, x_src, src2trg = prep_align(emb_tgt,emb_src,translate=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "53pfTQF4l_oR",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 272
        },
        "outputId": "ce704ce2-35e1-4a75-eecf-67fd439b9320"
      },
      "source": [
        "x_tgt_trans , x_src_norm = run_alignment(x_tgt,x_src,500,5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "*** Wasserstein Procrustes ***\n",
            "\n",
            "\n",
            "Computing initial mapping with convex relaxation...\n",
            "Done [049 sec]\n",
            "\n",
            "Computing mapping with Wasserstein Procrustes...\n",
            "epoch: 1  obj: 10.622\n",
            "epoch: 2  obj: 10.269\n",
            "epoch: 3  obj: 10.059\n",
            "epoch: 4  obj: 9.935\n",
            "epoch: 5  obj: 9.878\n",
            "Done [548 sec]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ymVNUd-Sxvpq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "8f57943b-5e77-44af-8157-78816b6e9aa8"
      },
      "source": [
        "similar=[]\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(x_tgt_trans[w_tgt.index(fr)],x_src_norm[w_src.index(de)]))\n",
        "np.mean(similar)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.5007437222840395"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BmnmBsd75O0m",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "fr_muse_align = dict(zip(w_tgt,x_tgt_trans))\n",
        "pickle.dump(fr_muse_align, open(path + '/embeddings/fr_muse_align.p', \"wb\" ))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v6OleY3fl5iq",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "dd4c4954-eb6a-4590-be01-97cfad4f54c0"
      },
      "source": [
        "similar = []\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(fr_muse_align[fr],de_muse_ext[de]))\n",
        "np.mean(similar)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.5007437339080246"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 44
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qyZvGSW_90ou",
        "colab_type": "text"
      },
      "source": [
        "## Unsupervised Wasserstein Procrustes\n",
        "\n",
        "using the facebook fasttext code : https://github.com/facebookresearch/fastText/blob/master/alignment/unsup_align.py  \n",
        "Paper: https://arxiv.org/abs/1805.11222"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FZmJ1xhIiLaw",
        "colab_type": "text"
      },
      "source": [
        "if the trained embeddings are aligned the cosine disteance between word pairs should be close to 1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hSscsOEN0L3h",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#de_own_ft = de_own_ft_w2v\n",
        "#fr_emb_ord = fr_emb_ord_w2v"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RN9gcr0LZbU5",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "515bfffa-061f-42b3-f1f4-d88c5b7fdba3"
      },
      "source": [
        "similar=[]\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(fr_emb_ord[fr],de_own_ft[de]))\n",
        "np.mean(similar)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.009146323"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 220
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BKX6MFOGwUb-",
        "colab_type": "text"
      },
      "source": [
        "Funciton are taken from facebook code"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "L1dqxgtrwUoa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def objective(X, Y, R, n=5000):\n",
        "    Xn, Yn = X[:n], Y[:n]\n",
        "    C = -np.dot(np.dot(Xn, R), Yn.T)\n",
        "    P = ot.sinkhorn(np.ones(n), np.ones(n), C, 0.025, stopThr=1e-3)\n",
        "    return 1000 * np.linalg.norm(np.dot(Xn, R) - np.dot(P, Yn)) / n\n",
        "\n",
        "\n",
        "def sqrt_eig(x):\n",
        "    U, s, VT = np.linalg.svd(x, full_matrices=False)\n",
        "    return np.dot(U, np.dot(np.diag(np.sqrt(s)), VT))\n",
        "\n",
        "\n",
        "def align(X, Y, R, lr=10., bsz=200, nepoch=5, niter=1000,\n",
        "          nmax=10000, reg=0.05, verbose=True):\n",
        "    for epoch in range(1, nepoch + 1):\n",
        "        for _it in range(1, niter + 1):\n",
        "            # sample mini-batch\n",
        "            xt = X[np.random.permutation(nmax)[:bsz], :]\n",
        "            yt = Y[np.random.permutation(nmax)[:bsz], :]\n",
        "            # compute OT on minibatch\n",
        "            C = -np.dot(np.dot(xt, R), yt.T)\n",
        "            P = ot.sinkhorn(np.ones(bsz), np.ones(bsz), C, reg, stopThr=1e-3)\n",
        "            # compute gradient\n",
        "            G = - np.dot(xt.T, np.dot(P, yt))\n",
        "            R -= lr / bsz * G\n",
        "            # project on orthogonal matrices\n",
        "            U, s, VT = np.linalg.svd(R)\n",
        "            R = np.dot(U, VT)\n",
        "        bsz *= 2\n",
        "        niter //= 4\n",
        "        if verbose:\n",
        "            print(\"epoch: %d  obj: %.3f\" % (epoch, objective(X, Y, R)))\n",
        "    return R\n",
        "\n",
        "# this function does not work\n",
        "def convex_init(X, Y, niter=100, reg=0.05, apply_sqrt=False):\n",
        "    n, d = X.shape\n",
        "    if apply_sqrt:\n",
        "        X, Y = sqrt_eig(X), sqrt_eig(Y)\n",
        "    K_X, K_Y = np.dot(X, X.T), np.dot(Y, Y.T)\n",
        "    K_Y *= np.linalg.norm(K_X) / np.linalg.norm(K_Y)\n",
        "    K2_X, K2_Y = np.dot(K_X, K_X), np.dot(K_Y, K_Y)\n",
        "    P = np.ones([n, n]) / float(n)\n",
        "    for it in range(1, niter + 1):\n",
        "        G = np.dot(P, K2_X) + np.dot(K2_Y, P) - 2 * np.dot(K_Y, np.dot(P, K_X))\n",
        "        #print(it,G,np.sum(np.isnan(G)))\n",
        "        q = ot.sinkhorn(np.ones(n), np.ones(n), G, reg, stopThr=1e-3)\n",
        "        #print(it,q,np.sum(np.isnan(q)))\n",
        "        alpha = 2.0 / float(2.0 + it)\n",
        "        P = alpha * q + (1.0 - alpha) * P\n",
        "        #print(it,sum(np.isnan(G)),sum(np.isnan(q)),sum(np.isnan(P)))\n",
        "    obj = np.linalg.norm(np.dot(P, K_X) - np.dot(K_Y, P))\n",
        "    #print(sum(np.isnan(X)),sum(np.isnan(Y)),sum(np.isnan(P)))\n",
        "    return procrustes(np.dot(P, X), Y).T\n",
        "\n",
        "def procrustes(X_src, Y_tgt):\n",
        "    U, s, V = np.linalg.svd(np.dot(Y_tgt.T, X_src))\n",
        "    return np.dot(U, V)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1hRldAglkPHb",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 333
        },
        "outputId": "81765004-a3ae-48af-8511-4468b77d37c2"
      },
      "source": [
        "# vocab french\n",
        "w_tgt = list(fr_emb_ord.keys())\n",
        "# embeddings french\n",
        "x_tgt = np.array([fr_emb_ord[vec] for vec in w_tgt])\n",
        "x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "# vocab geman\n",
        "w_src = list(de_own_ft.keys())\n",
        "# embeddings german\n",
        "x_src = np.array([de_own_ft[vec] for vec in w_src])\n",
        "x_src /= np.linalg.norm(x_src, axis=1)[:, np.newaxis] + 1e-8\n",
        "\n",
        "# dictionary of words to dictionary of index\n",
        "src2trg = dict(zip([w_src.index(repla(a)) for a in de_fr_dict.keys()],[w_tgt.index(repla(b)) for b in de_fr_dict.values()]))\n",
        "\n",
        "print(\"\\n*** Wasserstein Procrustes ***\\n\")\n",
        "import math\n",
        "import time\n",
        "import ot\n",
        "\n",
        "print(\"\\nComputing initial mapping with convex relaxation...\")\n",
        "t0 = time.time()\n",
        "R0 = convex_init(x_src[:2000], x_tgt[:2000], reg=0.1,niter=50, apply_sqrt=True)\n",
        "print(\"Done [%03d sec]\" % math.floor(time.time() - t0))\n",
        "\n",
        "print(\"\\nComputing mapping with Wasserstein Procrustes...\")\n",
        "t0 = time.time()\n",
        "R = align(x_src, x_tgt, R0.copy(),lr=.01,  bsz=200, nepoch=10, niter=1000, nmax=10000, reg=0.1, verbose=True)\n",
        "print(\"Done [%03d sec]\" % math.floor(time.time() - t0))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            "*** Wasserstein Procrustes ***\n",
            "\n",
            "\n",
            "Computing initial mapping with convex relaxation...\n",
            "Done [062 sec]\n",
            "\n",
            "Computing mapping with Wasserstein Procrustes...\n",
            "epoch: 1  obj: 11.504\n",
            "epoch: 2  obj: 11.503\n",
            "epoch: 3  obj: 11.503\n",
            "epoch: 4  obj: 11.503\n",
            "epoch: 5  obj: 11.503\n",
            "epoch: 6  obj: 11.503\n",
            "epoch: 7  obj: 11.503\n",
            "epoch: 8  obj: 11.503\n",
            "epoch: 9  obj: 11.503\n",
            "epoch: 10  obj: 11.503\n",
            "Done [084 sec]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FYni8bl-tSdO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if True:\n",
        "    x_src_ = x_src / np.linalg.norm(x_src, 2, 1).reshape([-1, 1])\n",
        "    #save_vectors(args.output_src, x_src, w_src)\n",
        "output_tgt=''\n",
        "if True:\n",
        "    x_tgt_ = x_tgt / np.linalg.norm(x_tgt, 2, 1).reshape([-1, 1])\n",
        "    #save_vectors(args.output_tgt, np.dot(x_tgt, R.T), w_tgt)\n",
        "x_tgt_ = np.dot(x_tgt_, R.T)\n",
        "#x_src = np.dot(x_src, R.T)\n",
        "similar=[]\n",
        "for de,fr in de_fr_dict.items():\n",
        "    similar.append(cosine_similarity(x_tgt_[w_tgt.index(fr)],x_src_[w_src.index(de)]))\n",
        "np.mean(similar)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jEW7PhffFu1J",
        "colab_type": "text"
      },
      "source": [
        "## take2\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HdIyVIfDFwob",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def smart_procrustes_align_gensim(base_embed, other_embed, words):\n",
        "    \"\"\"Procrustes align two gensim word2vec models (to allow for comparison between same word across models).\n",
        "    Code ported from HistWords <https://github.com/williamleif/histwords> by William Hamilton <wleif@stanford.edu>.\n",
        "      (With help from William. Thank you!)\n",
        "    First, intersect the vocabularies (see `intersection_align_gensim` documentation).\n",
        "    Then do the alignment on the other_embed model.\n",
        "    Replace the other_embed model's syn0 and syn0norm numpy matrices with the aligned version.\n",
        "    Return other_embed.\n",
        "    If `words` is set, intersect the two models' vocabulary with the vocabulary in words (see `intersection_align_gensim` documentation).\n",
        "    \"\"\"\n",
        "\n",
        "    # make sure vocabulary and indices are aligned\n",
        "    in_base_embed, in_other_embed = intersection_align_gensim(base_embed, other_embed, words=words)\n",
        "\n",
        "    # get the embedding matrices\n",
        "    base_vecs = np.array(list(in_base_embed.values()))\n",
        "    other_vecs = np.array(list(in_other_embed.values()))\n",
        "\n",
        "    # just a matrix dot product with numpy\n",
        "    m = other_vecs.T.dot(base_vecs) \n",
        "    # SVD method from numpy\n",
        "    u, _, v = np.linalg.svd(m)\n",
        "    # another matrix operation\n",
        "    ortho = u.dot(v) \n",
        "    # Replace original array with modified one\n",
        "    # i.e. multiplying the embedding matrix (syn0norm)by \"ortho\"\n",
        "    new_vec = np.array(list(other_embed.values())).dot(ortho)\n",
        "    new_vec /= np.linalg.norm(new_vec, axis=1)[:, np.newaxis] + 1e-8\n",
        "    other_embed =dict(zip(other_embed.keys(),new_vec))\n",
        "    return other_embed\n",
        "\t\n",
        "def intersection_align_gensim(m1,m2, words):\n",
        "    small_fr = {}\n",
        "    small_de = {}\n",
        "    for de,fr in words.items():\n",
        "        if de not in small_de.keys() and fr not in small_fr.keys() :\n",
        "            fr = repla(fr)\n",
        "            de = repla(de)\n",
        "            small_de[de] = m1[de]\n",
        "            small_fr[fr] = m2[fr]\n",
        "            if len(small_de.keys()) !=len(small_fr.keys()):\n",
        "                #print(m1[de],m2[fr])\n",
        "                break\n",
        "            #print(len(small_de.keys()),len(small_fr.keys()))\n",
        "\n",
        "    return (small_de,small_fr)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EuKSPS90XPyW",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "ceeb8a6c-5e0c-454e-9610-e243be01d819"
      },
      "source": [
        "de_fr_dict_ext = de_fr_dict\n",
        "de_fr_dict_ext.update(fr_de_db)\n",
        "len(de_fr_dict_ext.keys())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "269"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 185
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ok6lfSWvJil4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "fr_ft_alli = smart_procrustes_align_gensim(de_muse_ext,fr_muse_ext,de_fr_dict_ext)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dSvWTt_eOebS",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "ddeb8e75-70d7-42c3-a204-05a16c3a2425"
      },
      "source": [
        "similar = []\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(fr_muse_ext[fr],de_muse_ext[de]))\n",
        "np.mean(similar)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.4620076"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 187
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A5hlEa1eOqkZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "pickle.dump(fr_ft_alli, open(path + '/embeddings/fr_muse_alli.p', \"wb\" ))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Xhxsv-25nHhP",
        "colab_type": "text"
      },
      "source": [
        "## Align Supervised"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZSMzDE1T8AD8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def getknn(sc, x, y, k=10):\n",
        "    sidx = np.argpartition(sc, -k, axis=1)[:, -k:]\n",
        "    ytopk = y[sidx.flatten(), :]\n",
        "    ytopk = ytopk.reshape(sidx.shape[0], sidx.shape[1], y.shape[1])\n",
        "    f = np.sum(sc[np.arange(sc.shape[0])[:, None], sidx])\n",
        "    df = np.dot(ytopk.sum(1).T, x)\n",
        "    return f / k, df / k\n",
        "\n",
        "\n",
        "def rcsls(X_src, Y_tgt, Z_src, Z_tgt, R, knn=10):\n",
        "    X_trans = np.dot(X_src, R.T)\n",
        "    f = 2 * np.sum(X_trans * Y_tgt)\n",
        "    df = 2 * np.dot(Y_tgt.T, X_src)\n",
        "    fk0, dfk0 = getknn(np.dot(X_trans, Z_tgt.T), X_src, Z_tgt, knn)\n",
        "    fk1, dfk1 = getknn(np.dot(np.dot(Z_src, R.T), Y_tgt.T).T, Y_tgt, Z_src, knn)\n",
        "    f = f - fk0 -fk1\n",
        "    df = df - dfk0 - dfk1.T\n",
        "    return -f / X_src.shape[0], -df / X_src.shape[0]\n",
        "\n",
        "def proj_spectral(R):\n",
        "    U, s, V = np.linalg.svd(R)\n",
        "    s[s > 1] = 1\n",
        "    s[s < 0] = 0\n",
        "    return np.dot(U, np.dot(np.diag(s), V))\n",
        "\n",
        "def idx(words):\n",
        "    w2i = {}\n",
        "    for i, w in enumerate(words):\n",
        "        if w not in w2i:\n",
        "            w2i[w] = i\n",
        "    return w2i\n",
        "\n",
        "def select_vectors_from_pairs(x_src, y_tgt, pairs):\n",
        "    n = len(pairs)\n",
        "    d = x_src.shape[1]\n",
        "    x = np.zeros([n, d])\n",
        "    y = np.zeros([n, d])\n",
        "    for k, ij in enumerate(pairs):\n",
        "        i, j = ij\n",
        "        x[k, :] = x_src[i, :]\n",
        "        y[k, :] = y_tgt[j, :]\n",
        "    return x, y\n",
        "\n",
        "def compute_nn_accuracy(x_src, x_tgt, lexicon, bsz=100, lexicon_size=-1):\n",
        "    if lexicon_size < 0:\n",
        "        lexicon_size = len(lexicon)\n",
        "    idx_src = list(lexicon.keys())\n",
        "    acc = 0.0\n",
        "    x_src /= np.linalg.norm(x_src, axis=1)[:, np.newaxis] + 1e-8\n",
        "    x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "    for i in range(0, len(idx_src), bsz):\n",
        "        e = min(i + bsz, len(idx_src))\n",
        "        scores = np.dot(x_tgt, x_src[idx_src[i:e]].T)\n",
        "        pred = scores.argmax(axis=0)\n",
        "        for j in range(i, e):\n",
        "            if pred[j - i] == lexicon[idx_src[j]]:\n",
        "                acc += 1.0\n",
        "    return acc / lexicon_size\n",
        "\n",
        "def procrustes(X_src, Y_tgt):\n",
        "    U, s, V = np.linalg.svd(np.dot(Y_tgt.T, X_src))\n",
        "    return np.dot(U, V)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w8h5pPAQPH56",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "5599a006-1063-460b-b8c3-a8dddc506667"
      },
      "source": [
        "!wget https://dl.fbaipublicfiles.com/arrival/dictionaries/fr-de.0-5000.txt"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-06-19 18:21:19--  https://dl.fbaipublicfiles.com/arrival/dictionaries/fr-de.0-5000.txt\n",
            "Resolving dl.fbaipublicfiles.com (dl.fbaipublicfiles.com)... 104.22.74.142, 104.22.75.142, 172.67.9.4, ...\n",
            "Connecting to dl.fbaipublicfiles.com (dl.fbaipublicfiles.com)|104.22.74.142|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 86411 (84K) [text/x-c]\n",
            "Saving to: ‘fr-de.0-5000.txt’\n",
            "\n",
            "fr-de.0-5000.txt    100%[===================>]  84.39K   508KB/s    in 0.2s    \n",
            "\n",
            "2020-06-19 18:21:20 (508 KB/s) - ‘fr-de.0-5000.txt’ saved [86411/86411]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MmajqexiPZzm",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "ea03b5fa-84bd-4937-f0ff-b6bbe92fe721"
      },
      "source": [
        "filename = 'fr-de.0-5000.txt'\n",
        "words_src= de_muse.keys()\n",
        "words_tgt= fr_muse.keys()\n",
        "fr_de_db = {}\n",
        "with open(filename) as fh:\n",
        "    for line in fh:\n",
        "        fr_w = line.split()[0]\n",
        "        de_w = line.split()[1]\n",
        "        fr_w = repla(fr_w)\n",
        "        de_w = repla(de_w)\n",
        "        if fr_w in words_src and de_w in words_tgt and fr_w in fr_voc_count.keys() and de_w in de_voc_count.keys():\n",
        "            if (fr_voc_count[fr_w] > 100 and de_voc_count[de_w] > 150):\n",
        "                #print(de_w)\n",
        "                fr_de_db[fr_w] = de_w\n",
        "\n",
        "print(len(fr_de_db))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "186\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8nXPGz9lGeQR",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "129f0f6e-89e3-4413-8e3e-ecd019bb8804"
      },
      "source": [
        "de_fr_dict_ext = de_fr_dict\n",
        "de_fr_dict_ext.update(fr_de_db)\n",
        "len(de_fr_dict_ext.keys())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "269"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 160
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gCqEd5ZU0ba8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "de_s, fr_s = intersection_align_gensim(de_muse,fr_muse,de_fr_dict)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "48l4hwBy1C9r",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        },
        "outputId": "9b1412be-25b2-40aa-f8da-596a3f44d05b"
      },
      "source": [
        "###### MAIN ######\n",
        "# vocab french\n",
        "words_tgt = list(fr_s.keys())\n",
        "# embeddings french\n",
        "x_tgt = np.array([fr_s[vec] for vec in words_tgt])\n",
        "x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "# vocab geman\n",
        "words_src = list(de_s.keys())\n",
        "# embeddings german\n",
        "x_src = np.array([de_s[vec] for vec in words_src])\n",
        "x_src /= np.linalg.norm(x_src, axis=1)[:, np.newaxis] + 1e-8\n",
        "\n",
        "# dictionary of words to dictionary of index\n",
        "src2tgt = dict(zip([words_src.index(repla(a)) for a in de_fr_dict.keys()],[words_tgt.index(repla(b)) for b in de_fr_dict.values()]))\n",
        "lexicon_size = len(src2tgt)\n",
        "print(lexicon_size)\n",
        "\n",
        "# word --> vector indices\n",
        "idx_src = idx(words_src)\n",
        "idx_tgt = idx(words_tgt)\n",
        "\n",
        "# load word embeddings\n",
        "#words_tgt, x_tgt = load_vectors(params.tgt_emb, maxload=params.maxload, center=params.center)\n",
        "#words_src, x_src = load_vectors(params.src_emb, maxload=params.maxload, center=params.center)\n",
        "\n",
        "# load validation bilingual lexicon\n",
        "#src2tgt, lexicon_size = load_lexicon(params.dico_test, words_src, words_tgt)\n",
        "\n",
        "# word --> vector indices\n",
        "#idx_src = idx(words_src)\n",
        "#idx_tgt = idx(words_tgt)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "ValueError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-183-7fbcf1724018>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m \u001b[0;31m# dictionary of words to dictionary of index\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 14\u001b[0;31m \u001b[0msrc2tgt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mwords_src\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0ma\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mde_fr_dict\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mwords_tgt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mde_fr_dict\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     15\u001b[0m \u001b[0mlexicon_size\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc2tgt\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlexicon_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m<ipython-input-183-7fbcf1724018>\u001b[0m in \u001b[0;36m<listcomp>\u001b[0;34m(.0)\u001b[0m\n\u001b[1;32m     12\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     13\u001b[0m \u001b[0;31m# dictionary of words to dictionary of index\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 14\u001b[0;31m \u001b[0msrc2tgt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mwords_src\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0ma\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mde_fr_dict\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mwords_tgt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mb\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mb\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mde_fr_dict\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     15\u001b[0m \u001b[0mlexicon_size\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc2tgt\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlexicon_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mValueError\u001b[0m: 'normal' is not in list"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m-anywE3-1e0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# load train bilingual lexicon\n",
        "\n",
        "#pairs = list(zip([w_src.index(a) for a in fr_de_db.keys()],[w_tgt.index(b) for b in fr_de_db.values()]))\n",
        "pairs = list(zip([words_src.index(repla(a)) for a in de_fr_dict.keys()],[words_tgt.index(repla(b)) for b in de_fr_dict.values()]))\n",
        "# selecting training vector  pairs\n",
        "X_src, Y_tgt = select_vectors_from_pairs(x_src, x_tgt, pairs)\n",
        "\n",
        "# adding negatives for RCSLS\n",
        "maxneg=500\n",
        "Z_src = x_src[:maxneg, :]\n",
        "Z_tgt = x_tgt[:maxneg, :]\n",
        "print(X_src.shape,Y_tgt.shape)\n",
        "print(Z_src.shape,Z_tgt.shape)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "njeg-DJ8MnHf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def select_vectors_from_pairs(x_src, y_tgt, pairs):\n",
        "    n = len(pairs)\n",
        "    d = x_src.shape[1]\n",
        "    x = np.zeros([n, d])\n",
        "    y = np.zeros([n, d])\n",
        "    for k, ij in enumerate(pairs):\n",
        "        i, j = ij\n",
        "        print(k,i,j)\n",
        "        x[k, :] = x_src[i]\n",
        "        y[k, :] = y_tgt[j]\n",
        "    return x, y\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p1VMf8ODAiEl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import sys\n",
        "# initialization:\n",
        "R = procrustes(X_src, Y_tgt)\n",
        "#nnacc = compute_nn_accuracy(np.dot(x_src, R.T), x_tgt, src2tgt, lexicon_size=lexicon_size)\n",
        "#print(\"[init -- Procrustes] NN: %.4f\"%(nnacc))\n",
        "sys.stdout.flush()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-8EsILD9A0ul",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 945
        },
        "outputId": "3c08c024-8518-460a-8038-fc583b103331"
      },
      "source": [
        "# optimization\n",
        "fold, Rold = 0, []\n",
        "niter, lr = 50 , 1\n",
        "sgd = True\n",
        "batchsize = 10000\n",
        "knn=20\n",
        "reg = 0\n",
        "model = ''\n",
        "\n",
        "for it in range(0, niter + 1):\n",
        "    if lr < 1e-4:\n",
        "        break\n",
        "\n",
        "    if sgd:\n",
        "        indices = np.random.choice(X_src.shape[0], size=batchsize, replace=True)\n",
        "        f, df = rcsls(X_src[indices, :], Y_tgt[indices, :], Z_src, Z_tgt, R, knn)\n",
        "    else:\n",
        "        f, df = rcsls(X_src, Y_tgt, Z_src, Z_tgt, R, knn)\n",
        "\n",
        "    if reg > 0:\n",
        "        R *= (1 - lr * reg)\n",
        "    R -= lr * df\n",
        "\n",
        "    if model == \"spectral\":\n",
        "        R = proj_spectral(R)\n",
        "\n",
        "    print(\"[it=%d] f = %.4f\" % (it, f))\n",
        "    sys.stdout.flush()\n",
        "\n",
        "    if f > fold and it > 0 and not sgd:\n",
        "        lr /= 2\n",
        "        f, R = fold, Rold\n",
        "\n",
        "    fold, Rold = f, R\n",
        "\n",
        "    #if (it > 0 and it % 10 == 0) or it == niter:\n",
        "        #nnacc = compute_nn_accuracy(np.dot(x_src, R.T), x_tgt, src2tgt, lexicon_size=lexicon_size)\n",
        "        #print(\"[it=%d] NN = %.4f - Coverage = %.4f\" % (it, nnacc, len(src2tgt) / lexicon_size))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[it=0] f = -2.6214\n",
            "[it=1] f = -2.6457\n",
            "[it=2] f = -2.6768\n",
            "[it=3] f = -2.6986\n",
            "[it=4] f = -2.7206\n",
            "[it=5] f = -2.7448\n",
            "[it=6] f = -2.7754\n",
            "[it=7] f = -2.7966\n",
            "[it=8] f = -2.8267\n",
            "[it=9] f = -2.8573\n",
            "[it=10] f = -2.8795\n",
            "[it=11] f = -2.9011\n",
            "[it=12] f = -2.9347\n",
            "[it=13] f = -2.9526\n",
            "[it=14] f = -2.9829\n",
            "[it=15] f = -3.0087\n",
            "[it=16] f = -3.0316\n",
            "[it=17] f = -3.0587\n",
            "[it=18] f = -3.0917\n",
            "[it=19] f = -3.1132\n",
            "[it=20] f = -3.1388\n",
            "[it=21] f = -3.1625\n",
            "[it=22] f = -3.1896\n",
            "[it=23] f = -3.2143\n",
            "[it=24] f = -3.2431\n",
            "[it=25] f = -3.2705\n",
            "[it=26] f = -3.3010\n",
            "[it=27] f = -3.3191\n",
            "[it=28] f = -3.3503\n",
            "[it=29] f = -3.3687\n",
            "[it=30] f = -3.3929\n",
            "[it=31] f = -3.4231\n",
            "[it=32] f = -3.4533\n",
            "[it=33] f = -3.4768\n",
            "[it=34] f = -3.4997\n",
            "[it=35] f = -3.5358\n",
            "[it=36] f = -3.5499\n",
            "[it=37] f = -3.5765\n",
            "[it=38] f = -3.5972\n",
            "[it=39] f = -3.6292\n",
            "[it=40] f = -3.6617\n",
            "[it=41] f = -3.6801\n",
            "[it=42] f = -3.7077\n",
            "[it=43] f = -3.7396\n",
            "[it=44] f = -3.7532\n",
            "[it=45] f = -3.7852\n",
            "[it=46] f = -3.8137\n",
            "[it=47] f = -3.8316\n",
            "[it=48] f = -3.8642\n",
            "[it=49] f = -3.8822\n",
            "[it=50] f = -3.9067\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oPI9SYFe-Su8",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 225
        },
        "outputId": "9d833b02-79a9-4f01-df83-e3ee9b3bc1f8"
      },
      "source": [
        "x_tgt_t = np.dot(x_tgt, R.T)\n",
        "x_tgt_t /= np.linalg.norm(x_tgt_t, axis=1)[:, np.newaxis] + 1e-8\n",
        "\n",
        "\n",
        "nnacc = compute_nn_accuracy(x_src_t, x_tgt_t, src2tgt, lexicon_size=lexicon_size)\n",
        "print(\"[final] NN = %.4f - Coverage = %.4f\" % (nnacc, len(src2tgt) / lexicon_size))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-117-fb9cccf5d276>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m \u001b[0mnnacc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompute_nn_accuracy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_src_t\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mx_tgt_t\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msrc2tgt\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlexicon_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mlexicon_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0mprint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"[final] NN = %.4f - Coverage = %.4f\"\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mnnacc\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msrc2tgt\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m/\u001b[0m \u001b[0mlexicon_size\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'x_src_t' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NfIMgIhqU3Rz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "words_full = list(fr_muse.keys())\n",
        "# embeddings french\n",
        "x_full = np.array([fr_muse[vec] for vec in words_full])\n",
        "x_full /= np.linalg.norm(x_full, axis=1)[:, np.newaxis] + 1e-8\n",
        "\n",
        "x = np.dot(x_full, R.T)\n",
        "x /= np.linalg.norm(x, axis=1)[:, np.newaxis] + 1e-8\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CeXeT3ulJ4MA",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 225
        },
        "outputId": "8c020e32-f9f4-44bb-ddb8-93adb688c105"
      },
      "source": [
        "similar=[]\n",
        "for de,fr in de_fr_dict.items():\n",
        "    fr = repla(fr)\n",
        "    de = repla(de)\n",
        "    similar.append(cosine_similarity(x[words_tgt.index(fr)],de_muse[words_src.index(de)]))\n",
        "np.mean(similar)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "KeyError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-130-7fefd59166e1>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      3\u001b[0m     \u001b[0mfr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m     \u001b[0mde\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mrepla\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mde\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m     \u001b[0msimilar\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcosine_similarity\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mwords_tgt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mde_muse\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mwords_src\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mindex\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mde\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmean\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msimilar\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyError\u001b[0m: 0"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_mKGGPbsV3vi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if params.output != \"\":\n",
        "    print(\"Saving all aligned vectors at %s\" % params.output)\n",
        "    words_full, x_full = load_vectors(params.src_emb, maxload=-1, center=params.center, verbose=False)\n",
        "    x = np.dot(x_full, R.T)\n",
        "    x /= np.linalg.norm(x, axis=1)[:, np.newaxis] + 1e-8\n",
        "    save_vectors(params.output, x, words_full)\n",
        "    save_matrix(params.output + \"-mat\",  R)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ih6g_DSLkVUx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0NdNYqc0THej",
        "colab_type": "text"
      },
      "source": [
        "### Test new Embeddings"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "12gYRQI9TLoy",
        "colab_type": "text"
      },
      "source": [
        "#### In Vocab"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R-_zDNH84BNg",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 100
        },
        "outputId": "6a885bbf-13aa-4e0a-8f6c-2782e11cf5fb"
      },
      "source": [
        "similar('wurst',de_muse,5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('wurstbrot', 0.7768503),\n",
              " ('fleischwurst', 0.7504408),\n",
              " ('woerste', 0.7394904),\n",
              " ('wurstsuppe', 0.7378311),\n",
              " ('wurstbraet', 0.73757136)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XS0e0YaPSIss",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 100
        },
        "outputId": "bac3302e-3a6f-4580-9411-2ac0be078292"
      },
      "source": [
        "similar('wurst',de_own_ft,5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('rorwurst', 22.397615),\n",
              " ('briihwurst', 21.794773),\n",
              " ('wurstsulz', 21.249077),\n",
              " ('bqik', 20.898455),\n",
              " ('wurst', 20.343674)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XwmIpGxvSI9k",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 100
        },
        "outputId": "4190ba8c-f793-46c2-9266-89a143498438"
      },
      "source": [
        "similar('wurst',de_muse_ext,5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('kaesewurst', 0.8321126895329031),\n",
              " ('wurstaufschnitt', 0.8146291982446567),\n",
              " ('schinkenrotwurst', 0.7933772290902485),\n",
              " ('schinkenkrakauer', 0.7890030790415159),\n",
              " ('sommerwurst', 0.7848590018431179)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "koIfb1Jrdiyy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        },
        "outputId": "1844eb9c-eb9c-465d-db07-067ea949f0d7"
      },
      "source": [
        "word = 'pomme de terre'\n",
        "\n",
        "new_v = (fr_model['pomme'] +fr_model['de'] + fr_model['terre'] )/2\n",
        "print(new_v.shape)\n",
        "\n",
        "print(word,similar_v(new_v,de_model))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(300,)\n",
            "pomme de terre [('gedoenstet', 0.5515273), ('geschmort', 0.550975), ('lan', 0.5491309), ('liquer', 0.54181695), ('pochiert', 0.5408294)]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0tUfVFqnfUHp",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "a8cf2121-2908-48c8-8b5c-839341b2b529"
      },
      "source": [
        "fr = 'pomme'\n",
        "de = 'kartoffel'\n",
        "cosine_similarity(new_v ,de_model[de])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.3890639"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_1Q7tdCO3oir",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        },
        "outputId": "36a730ea-efc6-4235-a834-9f6c5ad7354a"
      },
      "source": [
        "word = 'backfertig'\n",
        "\n",
        "im_both_dict = shared_dict(de_own_ft,de_model)\n",
        "new_v = similar_both(word,de_own_ft,de_model,im_both_dict,10)\n",
        "print(word,similar_v(new_v,de_model))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "backfertig [('backzutat', 0.722348684554837), ('backzutaten', 0.7066722221245566), ('backform', 0.6963554129300749), ('kartoffelchips', 0.694915798786746), ('pizzateig', 0.690639124621061)]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1f60irExTSZS",
        "colab_type": "text"
      },
      "source": [
        "#### Out of Vocab"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IC5NbP-8D4qP",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "519e1e86-e413-4bfd-a5f7-ee2846d78985"
      },
      "source": [
        "ovv = pickle.load( open(path + '/embeddings/de_ooV.p', \"rb\" ) )\n",
        "\n",
        "ovv_to_emb = [w for w,cnt in  list(ovv.items()) if cnt>50]\n",
        "len(ovv_to_emb)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1652"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mawwx_I-Tzz7",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        },
        "outputId": "734794ae-da75-40b0-b4ec-bcc15a65cbdc"
      },
      "source": [
        "ovv_to_emb[11:15]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['vorratsschrank', 'instantgerichte', 'ffnen', 'kochhilfen']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u9tp7DmbTzwj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 278
        },
        "outputId": "a7ec81f0-2978-4422-9053-e67f5a05e092"
      },
      "source": [
        "similar('vorratsschrank',de_muse,5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "KeyError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyError\u001b[0m                                  Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-29-a646d3efe59b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0msimilar\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'vorratsschrank'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mde_muse\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;36m5\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m<ipython-input-8-681ac38a6cb3>\u001b[0m in \u001b[0;36msimilar\u001b[0;34m(word, embed, no)\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0msimilar\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mword\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0membed\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mno\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m5\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m     \u001b[0mmean\u001b[0m \u001b[0;34m=\u001b[0m  \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0membed\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mword\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      3\u001b[0m     \u001b[0mlimited\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0membed\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m     \u001b[0mdists\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlimited\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mmean\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m     \u001b[0mw2d\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mzip\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0membed\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeys\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0mlist\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdists\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyError\u001b[0m: 'vorratsschrank'"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Vo7g6LCQUHH0",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 100
        },
        "outputId": "388df33a-be19-485f-eab2-7aed04d92660"
      },
      "source": [
        "similar('instantgerichte',de_own_ft,5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('instantgerichte', 36.448486),\n",
              " ('kxey', 25.486507),\n",
              " ('bci', 24.919731),\n",
              " ('instantgelatine', 23.92138),\n",
              " ('pok', 23.387085)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zFBsMezPUHPy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 183
        },
        "outputId": "107d3a23-d116-4c56-928c-07452f4d4c67"
      },
      "source": [
        "similar('hund',de_muse_ext,10)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[('hundesnacks', 0.8330082140512194),\n",
              " ('hundesnack', 0.8293657795665598),\n",
              " ('hunde', 0.7618168828422787),\n",
              " ('hundes', 0.7617338501073201),\n",
              " ('katze', 0.7567193129512572),\n",
              " ('hunden', 0.707018349755564),\n",
              " ('hundebedarf', 0.7068084736316183),\n",
              " ('katzen', 0.6876648837708602),\n",
              " ('katzensnack', 0.6854314522589989),\n",
              " ('schwein', 0.6807642806359087)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w3_3Vd98UHqd",
        "colab_type": "text"
      },
      "source": [
        "Train"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qlg1rOj21kSt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "ovv = pickle.load( open(path + '/embeddings/de_ooV.p', \"rb\" ) )\n",
        "list(ovv.keys())[:50]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7EjoMOO1-EQI",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 67
        },
        "outputId": "cde63390-6b8d-40a4-a816-8bd0aa30c700"
      },
      "source": [
        "%%time\n",
        "from tqdm import tqdm\n",
        "\n",
        "for word in tqdm(ovv_to_emb):\n",
        "    if word in de_own_ft:\n",
        "        new_v = similar_both(word,de_own_ft,de_model,im_both_dict,10)\n",
        "        de_model[word] = new_v\n",
        "\n",
        "pickle.dump(de_model, open(path + '/embeddings/de_slim_embed_ext.p', \"wb\" ))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 1652/1652 [08:45<00:00,  3.15it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 10min 15s, sys: 7min 1s, total: 17min 17s\n",
            "Wall time: 8min 45s\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yjieskDCslWz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_tgt = list(de_own_ft.values())\n",
        "x_tgt /= np.linalg.norm(x_tgt, axis=1)[:, np.newaxis] + 1e-8\n",
        "de_own_ft = dict(zip(list(de_own_ft.keys()),x_tgt))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tZZAe-GUrdNN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "dist_milch = [[k,v] for k, v in sorted(w2d.items(), key=lambda item: item[1], reverse=True)]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k-4Y9YMDs4PE",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 100
        },
        "outputId": "d6045851-83a7-4063-ad5a-f5a4bbf2574a"
      },
      "source": [
        "dist_milch[:5]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[['milch', 0.99993813],\n",
              " ['buttermilch', 0.7569093],\n",
              " ['milcheis', 0.7549504],\n",
              " ['milchfett', 0.74762744],\n",
              " ['milchcreme', 0.7416733]]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 91
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oTWY1-YZorNS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "if not topn:\n",
        "    return dists\n",
        "best = matutils.argsort(dists, topn=topn + len(all_words), reverse=True)\n",
        "# ignore (don't return) words from the input\n",
        "result = [(self.index2word[sim], float(dists[sim])) for sim in best if sim not in all_words]\n",
        "return result[:topn]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1dxe8THpcEwd",
        "colab_type": "text"
      },
      "source": [
        "# Visualisation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T4GqvhxQNbs-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "%matplotlib inline\n",
        "import matplotlib.pyplot as plt\n",
        "plt.style.use('ggplot')\n",
        "\n",
        "from sklearn.decomposition import PCA"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rE2SSuHDYuO8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#pickle.dump(fr_muse_align, open(path + '/embeddings/fr_muse_align.p', \"wb\" ))\n",
        "\n",
        "fr_muse_align = pickle.load( open(path + '/embeddings/fr_muse_align.p', \"rb\" ) )\n",
        "fr_own_ft = pickle.load( open(path + '/embeddings/fr_embed_trained.p', \"rb\" ) )\n",
        "de_own_ft = pickle.load( open(path + '/embeddings/de_embed_trained.p', \"rb\" ) )"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xc8fbr9vO-tz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.manifold import TSNE\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "\n",
        "def tsne_plot(model_de,model_fr,model_fr2=None):\n",
        "    \"Creates and TSNE model and plots it\"\n",
        "\n",
        "    words_de =  [repla(w) for w in de_fr_dict.keys()]\n",
        "    words_fr =  [repla(w) for w in de_fr_dict.values()]\n",
        "    #words_fr2 =  [repla(w)+'_2' for w in de_fr_dict.values()]\n",
        "    embed_de =  [model_de[w] for w in words_de]\n",
        "    embed_fr =  [model_fr[w] for w in words_fr]\n",
        "    #embed_fr2 =  [model_fr2[w] for w in words_fr]\n",
        "    words = words_de + words_fr#+words_fr2\n",
        "    embed = embed_de+embed_fr#+embed_fr2\n",
        "    tsne_model = TSNE(perplexity=40, n_components=2, init='pca', n_iter=2500, random_state=23)\n",
        "    new_values = tsne_model.fit_transform(embed)\n",
        "\n",
        "    x = []\n",
        "    y = []\n",
        "    for value in new_values:\n",
        "        x.append(value[0])\n",
        "        y.append(value[1])\n",
        "        \n",
        "    plt.figure(figsize=(16, 16)) \n",
        "    for i in range(len(x)):\n",
        "        plt.scatter(x[i],y[i])\n",
        "        plt.annotate(words[i],\n",
        "                     xy=(x[i], y[i]),\n",
        "                     xytext=(5, 2),\n",
        "                     textcoords='offset points',\n",
        "                     ha='right',\n",
        "                     va='bottom')\n",
        "    plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ppFuJmLmQ2VT",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 918
        },
        "outputId": "a6b5eb76-b7ca-41f8-88c8-6724f5bb5b75"
      },
      "source": [
        "tsne_plot(de_own_ft,fr_ft_alli)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6AAAAOFCAYAAAB0pI80AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOzdeUBVdf7/8Seyy6a4kalNkkiLIpmhSYJ4cQOhDG3McWwat0m/pWk25tKU22Spqalpk6VmDWoqIi5hGNaPZNRGpWZcIlRccMNQ9u3+/qDuRIA798Ll9fiL+7mfc8773E/XeHE+53NsjEYjIiIiIiIiItWtnqULEBERERERkbpBAVRERERERETMQgFUREREREREzEIBVERERERERMxCAVRERERERETMQgFUREREREREzMLOQsfVs19ERERERESsm81vGywVQDlz5sxN9W/cuDEXL16spmrEXDSO1kNjaT00ltZB42g9NJbWQeNoPTSWt6Z58+aVtmsKroiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIiIiIiImIUCqIiIiIiIiJiFAqiIiIiIiIiYhQKoiIgAEB0dzeTJk29qm6ioKA4ePFhNFYmIiIi1UQAVERERERERs1AAFRGp4dLT0+nWrRtjx44lMDCQMWPGsHv3biIjI+natSv//ve/mTt3Lu+9955pm5CQENLT08nNzWXIkCEYDAZCQkKIiYkB4MCBA0RERGAwGOjatSvZ2dkAnDt3jsGDB9O1a1dmzJhh2l9iYiL9+vWjV69ejBgxgpycnAp1tmnThr///e8YDAbCw8O5cOFCNX8yIiIiUtsogIqI1ALHjx9n5MiR7N69mx9++IFNmzaxadMmpk2bxqJFi6rcbteuXXh5ebFz504SEhLo3r07hYWF/OUvf+GNN95g586dbNu2DScnJwC+//57li5dyhdffMHmzZs5ffo0mZmZLFiwgOjoaHbs2IGfnx/Lly+vcKzc3Fwefvhhdu7cSefOnVmzZk21fR4iIiJSOymAiojUAi1btuT++++nXr16+Pj4EBgYiI2NDb6+vqSnp1e5na+vL7t372bmzJkkJyfj7u5OamoqTZs2pUOHDgC4u7tjZ2cHQGBgIO7u7jg5OeHj48Pp06fZv38/R48eJTIyktDQUNatW8epU6cqHMvBwYHQ0FAA2rVrV2kfERERqdvsLF2AiIhcn6Ojo+nnevXq4eDgYPq5pKQEW1tbSktLTX0KCgoA8Pb2Zvv27SQkJDBnzhwCAwPp3bt3lcf5Zb+/7Lu4uBij0Ui3bt1YsmTJNWu0s7PDxsYGAFtbW4qLi2/+REVERMSq6QqoiIgVaNmyJSkpKQCkpKRw8uRJADIyMnB2duapp55i1KhRpKSk4O3tzfnz5zlw4AAAV69evWZY7NixI3v37iUtLQ0om2qbmppazWckIiIi1khXQEVErEDfvn1Zv3493bt3x9/fn9atWwNw+PBhZsyYgY2NDfb29syePRsHBweWLl3KlClTyM/Px83NjY8//rjKfTdq1Ij58+czevRoCgsLAZg4cSLe3t5mOTcRERGxHjZGo9ESxzWeOXPmpjZo3LgxFy9erKZyxFw0jtZDY2k9NJbWoapxnDBhAiNGjMDHx8cCVcmt0HfSOmgcrYfG8tY0b94cwOa37boCKiIit61e4SVcMuOxLb5KiZ0bOZ6hlDo0snRZArz99tuWLkFERMREAVRERG5LvcJLNDizArviTFObfX46PzV/TiHUjI4fP07fvn1p3749KSkp+Pj4sHDhQoYMGcLUqVM5d+4cb731FgD5+fkUFRWxbNkyJkyYAEBpaSmHDx/m9OnTrFmzhjVr1lBYWMi9997LwoULcXZ2tuTpiYiIldAiRCIicltcMuPLhU8Au+JMXDLjLVRR3ZWamsrQoUNJTEzEzc2NlStXmt7r2bMn8fHxxMfH88ADDzBq1Cj8/PxMbcHBwYwaNQqAPn36sHXrVnbu3Ml9993Hp59+aqlTEhERK6MroCIicltsi6/eVLtUn+bNm9OpUycA+vfvz4oVKyr0WbJkCU5OTjz77LOmtpiYGFJSUkxB88iRI8yZM4crV66Qk5NDUFCQWeoXERHrpwAqIiK3pcTO7abapfr88hzWql7v3r2bLVu28Nlnn5naDh8+zLx589iwYQO2trYAjBs3jg8++IAHH3yQ6Ohovvnmm+ovXkRE6gRNwRURkduS4xlKsZ1nubZiO09yPEMtVFHddfr0afbt2wfApk2bTFdDAU6dOsXkyZNZtmyZ6X7OrKwsRo8ezYIFC2jU6H/362ZnZ9OsWTOKiorYuHGjeU9CRESsmq6AiojIbSl1aMRPzZ/TKrg1gLe3NytXrmT8+PH4+PgwdOhQdu7cCcDatWu5fPkyzz33HABeXl7069ePU6dO8fLLL5v2ER8fz8svv0x4eDiNGjXC39+f7Oxsi5yPiIhYHz0HVMxK42g9NJbWQ2NpHbKzs4mIiCAhIcHSpcht0nfSOmgcrYfG8tboOaAiIiJyXbYFBbhlnMO2sJgSBzuuejWjxNHR0mWJiIiVUAAVERGxAr/73e9u++qnbUEBnqlp2BcWlTXkgH1OLpne9yqEiojIHaFFiERERAQAt4xz/wufP7MvLMIt45yFKhIREWujACoiIiIA2BYW31S7iIjIzVIAFREREQBKHCq/M6eqdhERkZulACoiIiIAXPVqRpGDfbm2Igd7rno1s1BFIiJibfQnTREREQGgxNGRTO97tQquiIhUGwVQERERMSlxdOSne1pZugwREbFSmoIrIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIiIiIiIiZqEAKiIiIiIiImahACoiIiIiIiJmoQAqIjVSbm4uQ4YMwWAwEBISQkxMDAEBAWRmZgJw8OBBoqKiAJg7dy7/93//R79+/ejatStr1qwBICkpif79+zNkyBAef/xxXnnlFUpLSwFITEykX79+9OrVixEjRpCTk2OZExURERGpQxRARaRG2rVrF15eXuzcuZOEhAS6d+9+zf7//e9/Wbt2LbGxscyfP5+MjAwADhw4wIwZM/jyyy85ceIEW7duJTMzkwULFhAdHc2OHTvw8/Nj+fLl5jgtERERkTrNztIFiIhUxtfXlzfeeIOZM2diMBgICAi4Zv9evXrh7OyMs7Mzjz32GAcOHMDd3Z0OHTpwzz33APDEE0/wr3/9C0dHR44ePUpkZCQARUVFdOzYsdrPSURERKSuUwAVkRrJ29ub7du3k5CQwJw5cwgMDMTOzs40hbagoKBcfxsbm0pfV9ZuNBrp1q0bS5YsqcYzEBEREZHf0hRcEamRMjIycHZ25qmnnmLUqFGkpKTQokULDh06BEBcXFy5/jt27CA/P5/MzEy++eYb/Pz8gLIpuCdPnqS0tJTNmzfz6KOP0rFjR/bu3UtaWhpQdr9pamqqeU9QREREpA7SFVARqZEOHz7MjBkzsLGxwd7entmzZ5Ofn8/48eN566236NKlS7n+999/PwMGDCAzM5OxY8fi5eXFjz/+iJ+fH5MnT+b48eM89thj9OnTh3r16jF//nxGjx5NYWEhABMnTsTb29sSpyoiIiJSZyiAikiNFBwcTHBwcIX2r7/+utL+DzzwAAsXLqzQ7urqyqpVqyq0BwYGsnXr1tuuU0RERERunAKoiNQ5WVlZ7Nmzh+zsbFxdXencuTMeHh6WLktERETE6imAikitN378+ErbH3vsMR577LFybVlZWWzcuJErV66Y2s6ePcuTTz6pECoiIiJSzbQIkYjUKXv27CkXPgGuXLnCnj17LFSRiIiISN2hACoidUp2dnal7Tk5OWauRERERKTuUQAVkTrF1dW10nYXFxczVyIiIiJS9yiAikid0rlzZ9zd3cu1ubu707lzZwtVJCIiIlJ3aBEiEalTPDw8ePLJJ9mzZw85OTm4uLhoFVwRERERM1EAFZE6x8PDg169elm6DBEREZE6R1NwRURERERExCwUQEVERERERMQsFEBFRERERETELBRARURERERExCwUQEVERERERMQsFEBFRERERETELBRARURERERExCwUQEVERERERMQsFEBFRERERETELOzuxE4GDhzYAPgH8BBgBJ5bu3btN3di3yIiIiIiImId7tQV0AXA9rVr1/oCfsB/79B+RURERERExErc9hXQgQMHegDdgGcB1q5dWwgU3u5+RURERCyhpKQEW1tbS5chImKVbIxG423tYODAgR2A5cB/KLv6uR94ce3atTm/6TcCGAGwdu3ajoWFN5dR7ezsKC4uvq1axfI0jtZDY2k9NJbWobaP46xZs/jkk09o0qQJLVq0wN/fn8jISF588UUuXrxI/fr1WbJkCW3atOH+++/nyJEjZGVlcdddd/H555/z+OOP06NHD9577z3++c9/4uLiwksvvQSAv78/Gzdu5He/+x2ffPIJixcvprCwkE6dOrFo0SJsbW3x9PRk2LBhJCQksGDBAsaOHQuUhdHvv/+egoICUlNTK9Tj6+t7xz+L2j6WUkbjaD00lrfGwcEBwOa37XfiHlA74GHg/9auXZs8cODABcBfgam/7rR27drllAVVAOPFixdv6iCNGzfmZreRmkfjaD00ltZDY2kdavM4HjhwgHXr1rF9+3aKi4vp1asXbdu2Zfjw4fz973+ndevWfPvtt/zlL39h3bp13HvvvSQlJXHy5EnatWvH559/TuvWrTlx4gQNGzYkNzcXGxsb0+dRUlLC5cuXOXv2LGvWrGH9+vXY29szadIkli9fzoABA8jJycHX15eJEycCsG3bNgCmT5/O448/zsWLF6us506rzWMp/6NxtB4ay1vTvHnzStvvRAA9BZxau3Zt8s+v11MWQEVERESua+/evfTq1QsnJycAQkNDyc/PZ//+/YwcOdLU75fZU48++ih79uwhPT2dMWPG8Mknn9ClSxf8/PyueZyvv/6alJQU+vbtC0B+fj6NGzcGwNbWlrCwsHL9Y2JiSElJ4dNPPyUnJ6fKekRE5MbddgBdu3ZtxsCBA9MHDhzYdu3atUeAHpRNxxURERG5JUajEXd3d+Lj4yu817lzZ1atWsW5c+eYMGECS5cuJSkpiYCAAKAsTJaWlpr6FxQUmPY5YMAAJk2aVGGfjo6O5e77PHz4MPPmzWPDhg2m/VVVj4iI3Lg7tQru/wFrBg4ceAjoAMy6Q/sVERERK9epUyfi4+PJz88nJyeHnTt34uzsTMuWLYmNjQXKwuP3338PQIcOHdi3bx82NjY4OTnx4IMP8vHHH5sCaMuWLUlJSQEgJSWFkydPAhAYGMiWLVtMU+kuX77MqVOnKtSTlZXF6NGjWbBgAY0aNQLAzc2tynpEROTG3ZHngK5du/YA8Mid2JeIiIjULR06dKBnz54YDAaaNGnC/fffj5ubG++++y6TJk1iwYIFFBcXExkZyYMPPoijoyPNmzfn4YcfBiAgIICYmBjuv/9+APr27cv69evp3r07/v7+tG7dGgAfHx8mTpzIoEGDMBqN2NnZMXPmTFq0aFGunh07dnDq1ClefvllU1t8fHyV9YiIyI277VVwb5HxzJkzN7WBbv61DhpH66GxtB4aS+tQ28cxJycHFxcX8vLy6N+/P3PmzKFdu3aWLssiavtYShmNo/XQWN6anxchqpZVcEVERERuy8SJEzl69CgFBQUMGDCgxoXP0gsZELMG40+Z2DTwhMjB1GviZemyRERqHQVQERERsbjFixdbuoQqlV7IwDh/GlzIAMAI8OMRSse9oRAqInKT7tQiRCIiIiLWKWaNKXya/HxFVEREbo4CqIiIiIVER0czefJkS5ch12H8KfOm2kVEpGoKoCIiYlXatGkDQEZGBsOHD7dwNeWVlJRYugS5BTYNPG+qXUREqqYAKiIiVsnLy4v333/frMf87LPPCAsLIzQ0lIkTJ1JSUkKbNm14/fXXMRgM7N+/n+joaAIDAwkLC2Pfvn2mbceOHcvUqVOJiIigS5cubNmyBYCkpCT++Mc/mvpNnjyZ6OhoAGbNmkVwcDAGg4FXXnkFgNjYWEJCQjAYDPTv39+MZ2/FIgfDb+/1bOJV1i4iIjdFAVRERGqtZcuWERISQkhISIWwmZ6eTkhICADh4eEcOXLE9F5UVBQHDx4kNzeXl156ibCwMHr27MmOHTuAsqmxw4YNY/DgwXTt2pUZM2Zct5Zjx46xefNmNm3aRHx8PLa2tmzYsIHc3Fz8/f3ZuXMn99xzD2+//TYxMTFs3LiRo0ePltvHuXPn2LRpEytXrmT27NnXPF5mZibbtm1j165d7Ny5k0mTJgHwzjvvsGbNGnbu3MmHH354/Q9RrqteEy9sxr2BTUAQtG2HTUAQNlqASETklmgVXBERqZUOHTrE2rVr2bJlC0ajkfDwcLp06VJp34iICGJjY2nbti3nzp3j3Llz+Pn5MXv2bLp27cq8efPIysoiLCyMxx9/HIDvv/+eHTt24ODgQLdu3fjTn/7E3XffXWU9X3/9NSkpKfTt2xeA/Px8GjdujK2tLWFhYQD8+9//pkuXLjRq1MhU148//mjaR+/evalXrx4+Pj5cuHDhmufv7u6Oo6Mj48ePx2Aw8Pvf/54rV67wyCOPMG7cOPr160efPn1u/AOVa6rXxAuGjbd0GTVaeno6Q4cOJSEhwdKliEgNpiugIiJSK/3rX/+id+/e1K9fHxcXF/r06UNycnKlffv160dcXBxQNkX1l0C4e/duFi9eTGhoKFFRURQUFHD69GkAAgMDcXd3x8nJCR8fH1N7VYxGIwMGDCA+Pp74+Hi++uorxo8fj6OjI7a2tjd0Tg4ODuX2B2BnZ2f6GaCgoMDUHhcXR1hYGDt37qRfv34AvPnmm0ycOJEzZ87Qp08fMjO1UI6IiNQcCqAiImL17rrrLho2bMh//vMfNm/eTEREBFAW8pYvX24KjXv37jUtYvTrMFivXj2Ki4uveYzAwEC2bNnCxYsXAbh8+TKnTp0q18ff3589e/aQmZlJUVGR6T7Pa7n77rs5evQoBQUFZGVl8fXXXwOQk5PD1atX6dGjB3/72984dOgQAMePH+fhhx/m5ZdfplGjRpw5c+YGPyWRO+fEiRP07NmTb7/9ln79+tGzZ08iIiL44YcfgLIFuaZPn07fvn0xGAysXr0aKJuG3r9/f0JDQwkJCTH9USkxMZF+/frRq1cvRowYQU5OjsXOTURuj6bgiohIrRQQEMC4ceMYM2YMRqOR7du3s3Dhwir7R0REsHTpUq5evcoDDzwAQFBQEB9++CEzZszAxsaG7777joceeuiW6vHx8WHixIkMGjQIo9GInZ0dM2fOLNenWbNmjB8/noiICDw8PHjwwQevu9+7776bfv36ERISQqtWrUz1ZWdn89xzz1FQUIDRaGTOnDkAzJgxg7S0NIxGI4GBgTd0DJE76YcffuD5559n/vz5tGrVio0bN2JnZ8fu3bt58803ef/99/n0009xc3Nj69atFBQU8MQTTxAUFMTWrVsJCgrixRdfpKSkhLy8PDIzM1mwYAHR0dHUr1+fxYsXs3z5csaNG2fpUxWRW6AAKiJiIVlZWWzcuJFnn3222o6RlJSEvb09nTp1qrZjWEq7du0YMGCAaTrtoEGDrhkew8LCmDZtGmPHjjW1jR07ltdeew2DwUBpaSktW7Zk1apVt1xTZGQkkZGR5dqOHTtW7vXTTz/N008/XWHbd955p8rtpkyZwpQpUyps88u0YoDGjRtz8eJF/vGPf9xS7SJ3wqVLl3juuef4xz/+YZq6PnbsWNLS0rCxsaGoqAgou6L53//+1/Tf8NWrV0lLS6NDhw6MHz+e4uJievXqxUMPPUR8fDxHjx41fbeKioro2LGjxc5RRG6Pza/vKzEj481OCfrlf6xSu2kcrYfG8vaZY8GOuXPn4uLiwqhRo6rso7G0Dg4uxSQeWU5e0U842zegXZMoXB2bWrosuQW19TuZnp7OoEGDaNmyJWFhYfzhD39g7NixtGvXjj//+c+kp6cTFRVFcnIyw4cPZ/DgwQQHB1fYT0ZGBl988QUfffQRI0aMwMPDg02bNrFkyRLzn9RtqK3jKBVpLG9N8+bNAWx+264roCIiFjJr1ixOnDhBaGgo3bp1A2DXrl3Y2NjwwgsvEBkZSVJSEnPnzsXd3Z3Dhw/Tr18/fH19+eCDD8jPz+eDDz7gd7/7HZ9//jkLFy6ksLCQhg0b8u6775Kfn8/q1auxtbXls88+Y8aMGdx333389a9/NS2o8/rrr2ul1EpkZBey5uBFLucW0bC+PYP9GuPl6nD9DS0ku+A8X6W+zZX8s6a2S7mpBN/zikKomJWDgwMffPABzzzzDC4uLly9ehUvr7LH1axdu9bULygoiFWrVtG1a1fs7e1JTU3lrrvuIjMzk7vuuovBgwdTWFhISkoKL774IpMnTyYtLY17772X3Nxczp49i7e3t6VOU0RugwKoiIiFvPrqqxw5coT4+Hji4uJYvXo18fHxZGZm0rdvXzp37gzAf/7zH7788ksaNGjAY489xqBBg4iLi+Mf//gHK1as4I033uDRRx8lNjYWGxsbPvnkE5YsWcJrr73GkCFDyl0BHT16NMOHD+fRRx/l9OnTPPPMMwqgv5GRXchrX6STkV30c0seRy/m8XqPljU2hKZcWF8ufALkFJ0n5cJ6urR43kJVSV1Vv359Vq5cyaBBgzAYDMyePZsFCxbQo0cPU59nnnmG9PR0evfujdFoxNPTkxUrVpCUlMR7772HnZ0dLi4uLFiwgEaNGjF//nxGjx5NYWEhABMnTlQAFamlFEBFRGqAf/3rXzzxxBPY2trSpEkTOnfuzMGDB3F1dcXPz49mzZoBcM899xAUFASAr68vSUlJAJw9e5a//OUvnD9/nsLCQlq1alXpcb766iuOHj1qep2dnU12dnY1n13tsubgxV+FzzIZ2UWsOXiR8V2bW6iqa8sr+qnS9vwq2kWqQ8uWLU23FHh4eLB161YAXnrpJVOfV155BShbWXrSpElMmjSp3D4GDhzIwIEDK+w7MDDQtD8Rqd0UQEVEarjfPg7kl9e/fjTI1KlTGTFiBD179iQpKYl58+ZVuq/S0lJiY2NxcnIytbm6upKfn1+NZ1C7XM4tqrw9r/L2msDZvkGl7U5VtIvUFrYFBbhlnMO2sJgSBzuuejWjxNHR0mWJyG3Qc0BFRCzExcXFdPUxICCAzZs3U1JSwqVLl0hOTqZDhw43vK8rV66Y7rNat25dpceA/z125Bfffffd7Z6G1WlY377ydufK22uCdk2icHe6q1ybi31T2jWJslBFIrfPtqAAz9Q06l/OwjEnh/qXs/BMTcO2oMDSpYnIbVAAFRGxEE9PTzp16kRISAj79+/n/vvvJzQ0lIEDBzJ58mSaNr3xxWPGjx/PyJEj6d27N56enqb20NBQtm/fTmhoKMnJyUyfPp2DBw9iMBgIDg42Pfxd/qdswaHyYdPLtWwhoprK1bEpke1n0cqjC03r308rjy5agEhqPbeMc9gXlp95YF9YhFvGOQtVJCJ3gh7DImalcfyfCRMmMGLECHx8fAgICGDbtm3lgkNNp7G0HhrLikyr4OYV0dC55q+CCxpHa6KxLNPo2I845uRUaC9wceFSm9YWqOjmaByth8by1ugxLCI1zNtvv23pEqQOqux+KqnIy9Whxi44JFJXlDjYQcX8WdYuIrWWpuCKVLP09HS6devGmDFjCAoKYvjw4eTl5REVFcXBgwfL9V21ahWhoaGEhobSuXNnoqLK7t9KTEykX79+9OrVixEjRpBTyV+ERa6nqvupyM2zdGkiIhVc9WpGkUP56fBFDvb6w5lILacAKmIGqampDB06lMTERNzc3Fi5cmWl/f74xz8SHx/P1q1bueuuuxgxYgSZmZksWLCA6OhoduzYgZ+fH8uXLzfzGYg1qOp+Ktsf0yxUkYhI1UocHcn0vpfchh4UuLiQ29CDTO97tQquSC2nOQwiZtC8eXM6deoEQP/+/VmxYsU1+0+bNo2uXbvSs2dP4uPjOXr0KJGRkQAUFRXRsWPHaq9ZrI9tYXGl7TZaUVJEaqgSR0d+uqfy5xqLSO2kACpiBjY2Ntd8/WvR0dGcOnWKmTNnAmA0GunWrRtLliyp1hrF+lV1P5VRVxNERETETDQFV8QMTp8+zb59+wDYtGmT6Wrobx06dIhly5axaNEi6tUr+3p27NiRvXv3kpZWNk0yNzeX1NRU8xQuVqWq+6lKWt9roYpERESkrlEAFTEDb29vVq5cSVBQEFlZWQwdOrTSfh9++CE//fQTAwYMIDQ0lAkTJtCoUSPmz5/P6NGjMRgMREREKIDKLanqfirqO1u6NBEREakjNAVXxAzs7OxYtGhRubb169ebfk5OTgZg/vz5lW4fGBjI1q1bq69AqTN0P5WIiIhYkgKoSA2Xk13CkZR88vNKcXKuR9t2Tri42lq6LBERERGRm6YAKlLNWrZsSUJCwi1tm5Ndwp4vs8nNMf7cUsLlS8V0DnZVCBURERGRWkf3gIrUYEdS8geVHE4AACAASURBVH8VPsvk5hg5kpJvoYpERERERG6dAqhIDZafV1p5e37l7SIiIiIiNZkCqEgN5uRc+VfUyUlfXRERERGpffRbrEgN1radE/VdbMq11XexoW07JwtVJCIiIiJy67QIkUgN5uJqS+dg17JVcPNLcXLSKrgiIiIiUnspgIrUcC6utjzcxcXSZYiIiIiI3DZNwRURERERERGzUAAVERERERERs1AAFREREREREbNQABURERERERGzUAAVERERERERs1AAFREREREREbNQABURERERERGzUAAVERERERERs1AAFREREREREbNQABURERERERGzUAAVERGRW5acnEz37t0JDQ0lLy+vyn4ffPABQUFBjBkzhoKCAp5++mlCQ0OJiYkxY7UiImJpdpYuQESkOkRHRxMUFISXl5elSxGxahs2bGDMmDE89dRT1+y3cuVK/vnPf9K8eXP2798PQHx8/A0fp7i4GDs7/doiIlLb6V9yEbE6JSUlrFu3Dl9fXwVQkTskPT2dwYMH0759e1JSUvDx8aFz585s2bKFxMREdu3axbvvvsvSpUuJjY2lsLCQ3r17M2HCBF555RVOnjzJkCFD6N+/P5988gmXLl0iNDSU999/nytXrvD666+Tk5ODp6cn8+fPp1mzZkRFRfHAAw+wd+9eIiMjGTVqlKU/BhERuU0KoCJSY+Xm5jJy5EjOnj1LaWkpL774Ip6enkyfPp2SkhL8/PyYPXs2jo6OBAQEEBERwe7duxkxYgQHDx5kzJgxODk5sXnzZo4dO1blL7j+/v4kJSWRlZXF3LlzCQgIsPSpi9RIqampzJ07l06dOvHSSy9RVFREaGgoBoOB8PBwEhMTSUtLIy4uDqPRyLPPPsuePXt48803+fLLL1m3bh2enp74+/vz3nvvsWrVKoqKinjhhRf48MMPadSoETExMbz55pvMmzcPgKKiIrZt22bhMxcRkTtFAVREaqxdu3bh5eXF6tWrAbhy5QohISFER0fj7e3NCy+8wKpVqxg+fDgADRs2ZMeOHQB8+umnTJ06FT8/P4qKipgyZUqVv+AWFxcTFxfHF198wbx584iOjrbMCYvUcM2bN6dTp04A9O/fnxUrVuDu7m56PzExkcTERHr27AmU/REpLS2Nzp07V7nP1NRUjhw5wu9//3sASktLadq0qen9iIiI6jgVERGxEAVQEamxfH19eeONN5g5cyYGgwFXV1datWqFt7c3AAMGDGDlypWmAFrVL6rX+wW3b9++ALRv355Tp05V5ymJ1Go2NjbXfG00GhkzZgxDhgy54X0ajUZ8fHyIjY2t9P369evffKEiIlJjaRVcEamxvL292b59O76+vsyZM8d0dbMqVf2i+ssvuPHx8cTHx/PFF1/w6aefmt53cHAAwNbWluLi4jt3AiJW5vTp0+zbtw+ATZs2ma6G/iI4OJjo6GhycnIAOHv2LBcvXrzmPr29vcnMzDTtt6ioiCNHjlRD9SIiUhMogIpIjZWRkYGzszNPPfUUo0aNYv/+/aSnp5OWlgbAZ599VuXUPhcXF7KzswH9gityp3h7e7Ny5UqCgoLIyspi6NCh5d4PCgriiSeeICIigh49ejBixAjT97AqDg4OLFu2jFmzZmEwGOjZs6fpuyoiItZHU3BFpMY6fPgwM2bMwMbGBnt7e2bPns2VK1cYOXKkaRGiqqb6DRw4kL/+9a+mRYiWLVvGtGnTuHLlCiUlJQwbNoy2bdua+YzEGiUlJZkW1PmtCRMmMGLECHx8fMq1R0dHc+jQIWbOnGmuMu8IOzs7Fi1aVK7tnXfeKfd62LBhDBs2rMK2ycnJpp8fe+wxHnvsMdPrhx56iA0bNlTYZv369bdbsoiI1DAKoCJSYwUHBxMcHFyh/fPPP6/Q9utfbgHCwsIICwszva7sF9x6hZfY8W4UtsX/oiTjv9TzDK2wH6l7SkpKsLW1vSP7evvtt+/IfuqSeoWXcMmMx7b4KiV2buR4hlLq0MjSZYmIyB2iKbgiUifVK7xEgzMrcM4+iEP+jzhnH6TBmRXUK7xk6dKkmj333HP07t2b7t278/HHHwPQpk0bXn/9dQwGA/v372f+/Pn07duXFi1aMHHiRM6ePcvw4cNJS0vj6aefxmAw0KtXL44fPw6UrfY6fPhwunXrxpgxYzAajQBERUVx8OBBoOyqZ2BgIGFhYeWmmF66dInhw4fTt29f+vbty969e837gdygli1bkpCQUK3H0PdSRMT66QqoiNRJLpnx2BVnlmuzK87EJTOeq16/t1BVYg5z586lYcOG5OXlERYWRt++fcnNzcXf35/XXnsNKAuk48aNo02bNuTl5ZGSksL7779PeHg4o0ePpk+fPuTn52M0Gjlz5gzfffcdCQkJeHl5ERkZyd69e3n00UdNxzx37hxvv/0227dvx83NjQEDBvDQQw8BMG3aNIYPH86jjz7K6dOneeaZZ0hMTLTIZ2Np1vK9HDt2rOnZqCIiUp4CqIjUSbbFV2+qXazHihUr2LZtGwBnzpwhLS0NW1tbzpw5Q0hICAB+fn4cO3aMvLw8kpKSaNasGbNmzeLq1assXryY1q1bm+4hfu211/D29qZBgwZMmDCBEydOMGbMGKZPn2465r///W+6dOlCo0ZlU0kjIiL48ccfAfjqq684evSoqW92djY5OTm4uLiY5fOoSfS9FBGxfgqgIlInldi53VS7WIekpCS++uorYmNjcXZ2JioqioKCAuzt7Vm3bh1btmwhLy8Pf39/PvroI0aOHMkzzzxDVlaWaR8RERHExsbStm1bzp07x+XLl3nggQdYsGABXbt2xdnZmTZt2jB9+nSaNGly3ZpKS0uJjY3FycmpOk+9VqjJ38vc3FxGjhzJ2bNnKS0t5cUXX+THH38kPj6e/Px8HnnkEd58881yz0Y9ePAgEyZMAMrG+fDhw5w+fdpSpyAiUiPoHlARqZNyPEMptvMs11Zs50mOZ6iFKqq7nn/+eQwGA8uXL6/2Y129ehUPDw+cnZ354Ycf+Pbbb4GyhYd69+5N/fr1sbOzw8HBwXRVMi4uDoB69epx11134ebmRlxcHAUFBWzYsMH0KKDdu3ezePFiYmJiWLx4MQUFBRQUFADg7+/Pnj17yMzMpKioiC1btphqCgoK4sMPPzS9/u6776r9c6ipavL3cteuXXh5ebFz504SEhLo3r07zz77LFu3biUhIYG8vDzi4+PLbePn52d6/nBwcDCjRo2yUPUiIjWHAqiI1EmlDo34qflz5Ln6UejUmjxXP35q/pxW2zSz8+fPc/DgQXbu3MmIESOq/XjBwcGUlJQQFBTErFmzePjhhyv08fDw4KGHHuLdd98lPz8fPz8/03sLFy5kw4YNpKen07NnTzZt2kTXrl0BMBqNLF++nMjISCZMmMDevXupX78+AM2aNWP8+PFERETwxBNP0KZNG9M+p0+fzsGDBzEYDAQHB7N69epq/hRqrpr8vfT19WX37t3MnDmT5ORk3N3dSUpKIjw8nB49epCUlFRuKvWvxcTEkJKSwquvvmrmqkVEah6bX1bqMzPjmTNnbmqDxo0bc/HixWoqR8xF42g9NJbW41bHsrIpiffeey+vv/46OTk5eHp6Mn/+fJo1a0ZUVBT+/v4kJSWRlZXF3LlzCQgIwGAwkJaWRuvWrZkxYwYBAQHVcIbXl5KSwrhx44iNjcVoNBIeHs7ChQt58sknOXbsGOnp6QwdOtS0CuxHH33E/v37+e6779i1axcAs2fPJjs72/Ts2u+++8600JA56DtZ/S5fvkxCQgKffPIJgYGBfPTRR2zdupW7776buXPnAjB+/PhyixAdPnyYkSNHsmHDBtM9wNejsbQOGkfrobG8Nc2bNwew+W27roCKiMgtqWxK4pQpU1i+fDnbt2/n6aef5s033zT1Ly4uJi4ujtdff5158+YB8OGHH3LPPfcQHx9vsfAJ0K5dOwYMGEBYWBjh4eEMGjTomuExLCyMmJgY+vXrZ2obO3YsRUVFGAwGunfvzpw5c657XNuCAhqcOEmjYz/S4MRJbH+esis1T0ZGBs7Ozjz11FOMGjWKlJQUADw9PcnJyTFN1f61rKwsRo8ezYIFC244fIqIWDstQiQiIrfE19eXN954g5kzZ2IwGPDw8ODIkSP8/vdlj8soLS2ladOmpv59+/YFoH379pw6dcoiNV/LyJEjGTlyZLm2Y8eOARWfgdmkSRNOnjxZrq+zs/MNhc5f2BYU4Jmahn1hUVlDDtjn5JLpfS8ljo63eBZSXQ4fPmy6um1vb8/s2bPZvn07PXr0oEmTJuWmav9ix44dnDp1ipdfftnU9tv7REVE6hoFUBERuSXe3t5s376dhIQE5syZQ9euXfHx8SE2NrbS/g4ODgDY2tpSXFxszlKrXU52CUdS8snPK8XJuR5t2znh4mp7zW3cMs79L3z+zL6wCLeMc/x0T6vqLFduQXBwMMHBweXa/Pz8eOWVVyr0feedd8jKymLPnj3MmjULV1dXOnfujIeHh5mqFRGpuRRARUTklmRkZNCgQQOeeuop3N3dWbVqFZmZmezbt49HHnmEoqIifvzxR9PzMq1VTnYJe77MJjfnlzUVSrh8qZjOwa7XDKG2hZWH8KrapfbIyspi48aNXLlyxdR29uxZnnzySYVQEanzFEBFROSWVDYl0dbWlmnTpnHlyhVKSkoYNmyY1QfQIyn5vwqfZXJzjBxJyefhLi5VblfiYAc5VbRLrbZnz55y4RPgypUr7Nmzh169elmoKhGRmkH/lxMRkVtS2ZREgA0bNlRoW79+velnT09PkpOTgYr3VtZG+XmllbfnV97+i6tezbDPyS03DbfIwZ6rXs3uaH1iftnZ2ZW25+RU8hcHEZE6RgFURETMKrvgPCkX1pNX9BPO9g1o1yQKV8em19+whnJyrgeUVGx3uvZC8yWOjmR634tbxjlsC4spcbDjqlczLUBkBVxdXSttd3Gp+oq4iEhdoQAqIiJmk11wni9PvElO0XlT26XcVILveaXWhtC27Zy4fKm43DTc+i42tG3ndN1tSxwdteCQFercuTNnz54tNw3X3d2dzp07W7AqEZGaQQFURETMJuXC+nLhEyCnqOyKaJcWz1uoqtvj4mpL52DXslVw80txcrqxVXDFenl4ePDkk0+yZ88ecnJycHFx0Sq4IiI/UwAVERGzySv6qdL2/CraawsXV9trLjgkdY+Hh4cWHBIRqcS1b1ARERG5g5ztG1Ta7lRFu4iIiFgXBVARETGbdk2icLEvf6+ni31T2jWJslBFIiIiYk6agisiImbj6tiU4HteIeXCevKLfsLJClbBFRERkRunACoiImbl6ti01i44JCIiIrdHU3BFRERERETELBRARUR+NmTIELKysq7ZZ+HChWaqRkRERMT6KICKSJ1nNBopLS1l9erV131O36JFi8xUlYiIiIj1UQAVEasxa9YsPvroI9PruXPn8s477zBw4EB69epFjx492LFjBwDp6ek8/vjjvPDCC4SEhHDmzBkCAgLIzMwE4LPPPiMsLIzQ0FAmTpxISUkJs2bNIj8/n9DQUMaMGVNlv5KSEsaOHUtISAg9evRg+fLlZv8sRERERGoiLUIkIlYjIiKC1157jWeffRaA2NhY1qxZw5///Gfc3NzIzMykX79+9OzZE4C0tDTeeecdOnbsWG4/x44dY/PmzWzatAl7e3smTZrEhg0bePXVV/nwww+Jj4+/Zr+2bduSkZFBQkICwHWn9YqIiIjUFQqgImI1HnroIS5evEhGRgaXLl3Cw8ODpk2b8re//Y3k5GRsbGzIyMjgwoULALRo0aJC+AT4+uuvSUlJoW/fvgDk5+fTuHHjG+4XGhrKyZMnmTJlCj169CAoKKgaz1pERESk9lAAFRGrEh4eTlxcHOfPnyciIoINGzZw6dIltm3bhr29PQEBARQUFABQv379SvdhNBoZMGAAkyZNuuaxrtUvPj6eL7/8ktWrVxMbG8u8efNu/+REREREajndAyoiViUiIoKYmBji4uIIDw/n6tWrNG7cGHt7e/7f//t/nDp16rr7CAwMZMuWLVy8eBGAy5cvm7azt7enqKjomv0yMzMpLS0lLCyMiRMnkpKSUk1nKyJSO126dInw8HB69uxJcnJypX2io6PJyMioch9vvfUWu3fvrvL96OhoJk+efNu1isidpSugImJV2rZtS05ODl5eXjRr1oz+/fszdOhQevToQfv27bnvvvuuuw8fHx8mTpzIoEGDMBqN2NnZMXPmTFq0aMHgwYMxGAy0a9eOd999t9J+Tk5OvPTSS5SWlgJc90qqiEhd8/XXX+Pr68vbb79dZZ9169bh6+uLl5dXhfdKSkp4+eWXq7NEEakmNkaj0RLHNZ45c+amNmjcuLHpKoPUXhpH66GxtB4aS+ugcbQetXUsn3vuOc6cOUNBQQF//vOf+cMf/kCbNm04duwYAFu2bGHnzp0MGzaMP/3pT+Tn5+Pl5UVMTAyvvvoqhw4dwsbGhqeffprmzZszbtw4vLy8cHJyYvPmzQQHBxMREcHu3bt5/vnn2bVrFwaDgfDwcA4cOMC0adPIzc3F0dGR6Oho4uLiiI+PJy8vj+PHj9OnTx+mTJkCQGJiIm+//TaFhYXcc889zJ8/HxcXlzv6edTWcZSKNJa3pnnz5gA2v23XFVARkTsou+A8KRfWk1f0E872DWjXJApXx6aWLktEpNrNnTuXhg0bkpeXR1hYmGmBtt966KGHmDBhAocOHWLmzJkcOnSowsrhHh4efPTRR0ydOhU/Pz/Ttg0bNjQ9TmvXrl0AFBYW8pe//IWlS5fSoUMHrl69ipOTEwDff/89O3bswMHBgW7duvGnP/0JZ2dnFixYQHR0NPXr12fx4sUsX76ccePGVefHIyI/UwAVEblDsgvO8+WJN8kpOm9qu5SbSvA9ryiEiojVW7FiBdu2bQPgzJkzpKWl3dB2rVq1uuGVwyMiIiq0paam0rRpUzp06ACAm5ub6b3AwEDc3d2BstsrTp8+TVZWFkePHiUyMhKAoqKiSldEF5HqoQAqInKHpFxYXy58AuQUlV0R7dLieQtVJSJS/ZKSkvjqq6+IjY3F2dmZqKgoCgoKsLH53+y7X1Yg/60GDRrc8MrhVa1eXhUHBwfTz/Xq1aO4uBij0Ui3bt1YsmTJTe1LRO4MrYIrInKH5BX9VGl7fhXtIiLW4urVq3h4eODs7MwPP/zAt99+C0CTJk04duwYpaWlbN++vdJtq1o53MXFhezs7Ose29vbm/Pnz3PgwAEAsrOzKS4urrJ/x44d2bt3r+kKbW5uLqmpqTd1viJy63QFVETkDnG2b1Bpu1MV7SK10YQJExgxYgQ+Pj6WLkVqkODgYFavXk1QUBDe3t48/PDDQNkq4EOHDsXT0xM/Pz9ycnIqbHv27NlKVw4fOHAgf/3rX02LEFXFwcGBpUuXMmXKFPLz83FyciI6OrrK/o0aNWL+/PmMHj2awsJCACZOnIi3t/ctn7+I3DitgitmpXG0HhrLiiq7B9TFvmmNvwdUY2kdauI4Go1GjEYj9eppwtXNqIljaU1KL2RAzBqMP2Vi08ATIgdTr0nFR73cLo2j9dBY3hqtgisiUs1cHcvCZsqF9eQX/YSTVsGVWiA9PZ3BgwfTvn17UlJS8PHxYeHChezbt4/p06dTUlKCn58fs2fPxtHRkaioKKZOncrBgwc5ceIEU6dOBSA6OppDhw4xatQonnnmGfz9/UlJSTHd0xcbG0thYSG9e/dmwoQJFj5rqatKL2RgnD8NLmQAYAT48Qil496olhAqIhXpT5IiIneQq2NTurR4nu73vkqXFs8rfEqtkJqaytChQ0lMTMTNzY1ly5Yxbtw4li5dyhdffEFxcTGrVq0qt01YWFi5e/piY2NNq4qmpaUxdOhQdu3aRWpqKmlpacTFxfH5559z6NAh9uzZY9bzEzGJWWMKnyY/XxEVEfNQABUREanjmjdvTqdOnQDo378/X3/9Na1atTLdEzdgwACSk5PLbdOoUSNatWrF/v37yczM5IcffjDto0WLFqbHWiQmJpKYmEjPnj3p1auXKZBKeenp6YSEhNxw/6SkJPbu3Wt6vX37do4ePVodpVkV40+ZN9UuIneepuCKiIjUcb9+VAaAh4cHly9fvu52kZGRxMbGct9999G7d2/Tfn79qAyj0ciYMWMYMmTInS26jvvmm29wcXExhf7t27djMBhuanGo4uJi7Ozq1q+CNg08qWz1E5sGnmavRaSu0hVQERGROu706dPs27cPgE2bNtG+fXvS09NNVyo/++wzOnfuXGG73r178/nnn7Np0ybT9NvfCg4OJjo62rT66dmzZ7WYRxWKi4sZM2YM7du3Z/jw4eTl5REQEEBmZtnVuYMHDxIVFUV6ejqrV6/m/fffJzQ0lG+++Yb4+HhmzJhBaGgox48f5/jx4wwePJjevXvz5JNP8sMPPwAwduxYXnnlFcLDw5kxY4YlT9cyIgfDb+/1bOJV1i4iZlG3/uwlIiIiFXh7e7Ny5UrGjx+Pj48P06dP5+GHH2bkyJGmRYgqu4LZoEED7rvvPo4dO4a/v3+l+w4KCuLYsWNEREQAZVdHFy1aROPGjav1nGqj1NRU5s6dS58+ffjjH//IypUrK+3XsmVLhgwZgouLC6NGjQIgNDQUg8FAeHg4UPYIk7///e+0bt2ab7/9lkmTJrFu3Tqg7I8AMTEx2NramufEapB6TbwoHfeGWVbBFZHKKYCKiIjUcXZ2dixatKhc2+OPP87nn39eoe/69evLvf7t4kQtW7YkISGhXNuwYcMYNmzYHarWev32XtwVK1bc0n5ycnLYv38/I0eONLX98rxLgPDw8DoZPn9Rr4kXDBtv6TJE6iwFUBEREakW2QXnSbmwnryin3DWY4mu67f34trY2GBnZ0dpaSkABQUFN7Sf0tJS3N3diY+Pr/T9X9+jKyJibroHVEREpA6r7IrljXjrrbfYvXt3le9nF5znyxNvcjLrGy7k/peTWd/w5Yk3yS44fzvlWrXf3ovbqVMnWrRowaFDhwCIi4sz9XVxcSE7O9v02tXV1XSfrZubGy1btiQ2NhYoWwjq+++/N9dpiIhckwKoiIiF5ObmMmTIEAwGAyEhIcTExPDVV1/Rs2dPevTowUsvvWS64jFr1iyCg4MxGAy88cYbQNlzF0NCQjAYDPTv3x+AkpISpk+fTt++fTEYDKxevdpi5yfW7eWXX6Zbt25Vvp9yYT05ReXDZk5R2RVRqdwv9+K2b9+erKwshg4dyksvvcS0adPo06dPuWmzoaGhbN++ndDQUJKTk4mMjGTp0qX07NmT48eP8+677/LPf/4Tg8FA9+7dK51OLSJiCTZGY2WLUVc745kzZ25qg8aNG2vVPCtwJ8YxOjqaQ4cOMXPmzArvDRkyhHfffRcPD48qt4+KimLq1Kn4+fndVh11nb6Tty8uLo4vv/ySt956C4ArV64QEhJCdHQ03t7evPDCC7Rr146nnnqKyMhIdu/ejY2NDVlZWXh4eNCjRw8+/vhj7rrrLlPbxx9/zMWLFxk7diwFBQU88cQTLFu2jFatWlVZR3WOZVZWFhs3buTZZ5+9rf3MnTu33IIrUtGdGMfPPvuMFStWUFhYiL+/P7Nnz8bX15dnnnmG3bt306RJE5YuXUqjRo0YO3asadGbAwcOMG3aNHJzc3F0dCQ6Opo9Ge/w4aI4Ug9dpriolK79WvBY+N3Y57bi49n/5erVq5SUlDB79mwCAgLu0KdgHfTvq3XQOFoPjeWtad68OYDNb9t1BVSsyurVq68ZPkVqEl9fX3bv3s3MmTNJTk4mPT2dVq1a4e3tDcCAAQNITk7G3d0dR0dHxo8fz9atW3F2dgbgkUceYdy4caxZs4aSkhIAEhMTWb9+PaGhoYSHh3P58mXTozQs4cqVKxUWqbkeo9FouudNzOfYsWNs3ryZTZs2ER8fj62tLRs2bCA3Nxc/v//P3p0HRFX1DRz/wrDKKqISSmW4kQpqIiYICgO4k3tpvpRpWFGipuZWuZaauz6iT+6ZueSS4hKIS4VL+eRSKW6ouGDpIMKww7x/8HAfRxY3dn+fv+aeuffcc+Yywz33nPM7buzfv59XX32VOXPm6B2XmZnJe++9x+TJk4mKiuK7777DzMyMmN2XMbdQMXyRO8MXunNk9w3u3Ezj1+h4fHx8iIyMJDIykiZNmpRTjZ8dqowMbK9cpcb5S9heuYrqEeeSCiFEaZAGqKgwChuOeOLECbp3745araZLly7KfJdbt24xYMAAPD099dYxy18vLT4+7wZn1KhRdOjQgTfeeIO0tDS98+Xm5hIWFsaMGTOKHLYYExND7969GTJkCN7e3oSGhlJOowYKtWfPHs6dO1cieW3YsIHx48c/dT5JSUmsWrXq6Qv0DHB2dmbPnj00btyYmTNnsnfv3kL3MzIyIiIigi5duhAVFcWAAXnr1c2YMYPRo0dz48YNOnXqpKwVOHXqVOXm/siRI/j4+DxSeeLj4/H29iYsLAwvLy9CQ0M5dOgQQUFBeHp68vvvv5OYmMigQYOUnq+//voLyOuhDA8PV/Ly9fUlPj6e6dOnc+XKFfz9/ZkyZQparZa+ffsSGBiIn5+fUuf4+HjatWvHRx99hK+vLzdu3GD+/Pl4eXnx2muvcfHiRSXvotY3FE/n559/5vTp03Tu3Bl/f39+/vlnrl69iqGhobKESs+ePTl27JjecRcvXqRWrVo0b94cyJt/aGRkxOUTmRzfd5uvhh5j3ke/kXovi5RbJgS0fYONGzcye/Zszpw5g6Wl5VOXPSwsjJ07dz51PlWRKiMDu4txVEtMwlSrpVpiEnYX46QRKoQoNyUWBbdv374q4Dfg+saNG7uWVL7iEdByJgAAIABJREFU2bF//34cHByUxt+9e/cIDAxkyZIlNG/enOTkZMzMzAD4888/2bt3LyYmJnh7e/P2229Tp04dvfzi4uJYvHgxs2bNIiQkhF27dtGrVy/gf4t9N2rUiGHDhvHNN99gZWXFrl27lGGL+Tftf/zxB9HR0Tg4OBAUFMSvv/5K69aty/CTKdqePXtQq9U0bNiwwHvZ2dkYGZV9oOv8Hq+nHXL5LEhISMDW1pZevXphbW3NqlWriI+PJy4ujnr16vH999/Tpk0btFotaWlp+Pn54e7uzquvvgrkNcRatmxJy5Yt2b9/Pzdu3MDHx4c1a9bg6emJsbExFy9e5LnnnnvkqJeXL19m6dKlzJkzh86dO7Nt2za2bdvGjz/+yMKFC3F0dKRp06asWLGCn3/+mWHDhhUZaRNg3LhxxMbGKvtkZ2ezfPlyrKys0Gg0dOvWjYCAACDvOztv3jxeeeUVTp06xQ8//EBkZCTZ2dl07NgRV1dXAEaPHl3k+obiyel0Ovr06cPYsWP10ufNm6e3/WCk1qKoDE2ZPvVLbF1ukJ51F7P7ouB+/30L9u3bx/Dhw3n33Xfp06dPidVD6LNKuIVxZpZemnFmFlYJt7j7QtFD84UQorSU5N3pMOAMYF2CeYpnSOPGjZk8eTLTpk1DrVZjbW1d4Kl6Pi8vL6yt8/7UGjZsyPXr1ws0QJ2cnGjatCkArq6uxMfHK++NGTOGbt26MWzYMCBv2OKZM2eUCIPJycnExcVhbGxM8+bN88ew06RJE+Lj40u1ATp37ly2bNlCjRo1cHR0xNXVlY4dOzJ+/Hju3LmDubk5s2bNIjExUenhmj9/Pv/+97/5+OOPefnll/n1118JCgqiSZMmTJkyRVlI/osvvsDU1LTQ+Vr3i4qKYv78+axevZrJkydjZWXFyZMn+eeffxg/fryy0PmSJUvYsWMHmZmZdOzYkY8//livx8vb25uJEyeW2mdV2Z09e5apU6diYGCAsbExX3zxBffu3SMkJES5ZgMHDuTu3bsMGjSIjIwMdDodn332GZDX0xkXF4dOp8PLy4smTZrw8ssvEx8fT8eOHdHpdNjZ2T3WWoJOTk64uLgAed8tLy8vDAwMaNy4MfHx8Vy7do1///vfQN73MDExkeTk5EfOX6fT8eWXX3L06FEMDAxISEjgn3/+AaBu3bq88sorABw9epSOHTsqw439/f2Bh69vKJ6cl5cXb7/9NkOGDMHe3p7ExES0Wi25ublEREQQFBTE1q1bC/z+OTs78/fff3PixAmaN29OSkoKZmZm+Pj4sGn9DyxdulR5GGKYY8m1a9d47rnnGDBgAJmZmZw+fbrQBmhqaiohISHcvHmT3Nxchg0bRr169Zg0aRJarRY7Ozvmzp1L7dq1y+ojqpRUmdmPlS6eHUePHuWTTz7ByMiIH374Qfm9fRLx8fEEBwc/UURt8ewpkQZo37596wJdgGnAiJLIUzx78ocjRkdHM3PmTDw9PYvc18TERHltaGhIdnbBf6SmpqbKa5VKRXp6urLdqlUrYmJiCAkJUXpVp06dSvv27fXyiImJ0TuXSqUq9Fwl5cSJE+zatUvp9QkMDMTV1bXIHh9/f39lKGS+rKwsdu/eTXp6Ol5eXnoBbdasWUNwcDDvvfdeoT3LALt372bZsmWsXbsWW1tbIG/I87Zt27hw4QJvv/02Xbt2JTIykri4OCIiItDpdLz11lscOXKkQI+XKFr79u0L/M0BBaJV1q5dW2/5hXxff/213nZCSibrTt5G26o/nb2DGeBmj4OlSYHjinP/98bQ0FD5+zc0NCQnJ6fIXnWVSqU3b7Oo9Qq3bNnCnTt32L17N8bGxnh4eCj7Pkov7cPWNxRPrmHDhowePZo33ngDnU6HkZER06ZNo1q1avz+++/Mnz+fGjVq6A21hrzf4yVLljBhwgTS09MxMzNjw4YN9O/fv9CHITExMYSHh2NkZISFhQXz588vtDyFjYp58803WblyJTVq1GD79u3MmDGjwJxUoS/HxAi0RaSLZ9qWLVsIDQ1VRocJUVZK6tdnHjAasCpqh759+74LvAuwceNG7O3tH+sERkZGj32MqHiKu443btzAycmJkJAQ6taty9KlS7l9+zaXL1+mVatWJCcnY25ujpWVFWZmZko+JiYm2NjYYG9vj0qlws7ODhMTE1QqlbJPtWrV0Ol02NvbY2xszHvvvcdPP/3ERx99xMaNG+nSpQvfffcdQUFBGBsbc+7cOerUqYONjQ0mJiZKPmZmZlhZWZXa3+Jff/1Fjx49qFu3LgDdu3dHpVJx/PhxPvjgA2W/jIwM7O3tMTMzw9raWimPsbExAwcOxN7enlOnTuHs7KxElxw8eDDh4eF069aNOnXqoFarAZRjraysOHLkCH/99RcRERFKD7OZmRldunShVq1a1KpVi9u3b2Nvb090dDQ///wznTt3BiAlJYV//vmHpk2b6n32omzcSEpj8oHLXE/Kf9CSxsXETOb1aIKjTfFPtfO/lykpKXrX7v6/r/z3fHx82Lt3L+PGjePgwYPUqlWLevXq8fLLL7Nr1y7s7e35/fffuXr1KtWrV8fKyoq0tDQlz9zcXOrWrctzzz3HgQMHuHbtGtWrVwfQO3fHjh0ZPHgwn3/+OdnZ2URHRzN48GDq1avHSy+9xMGDB+nVqxc6nY7Tp08rw3OfZSXxf/Kdd97hnXfe0UszMDBg0aJFBfb95ptvlNdqtVr5Tbnf7NmzC6S9//77vP/++w8ty6uvvsq0adOU4eDVq1fn3LlzvPnmm0DekkMODg6F/hZWdiV6z1PNAt3vJzG4Lw6CztwcIxcX7Ks9eY+XeLiKcu96+fJlunXrRsuWLfn99995+eWX8fLyIiIigp9++omYmBgWL15Mr169uHv3LllZWXz++ed0796dy5cv0717dzw9PTl8+DCOjo58//33mJub85///Id3330XyPsNqMr/+yvKtawqnroB2rdv367A3xs3bjzet2/f9kXtt3HjxmXAsv9u6h43lLGEP64airuOMTExBYYj6nQ6QkND9Z6qJycnk56eruSTmZlJUlISt2/fJicnB41Gg1arJScnR9knNTWV1NRUbt++TVZWFnfv3uXNN98kISGB/v37s2jRIs6cOUOrVq30ntQnJSWRmZmp5JOenk5ycnKp/S1qtVqlnABpaWmkpKRgbW3N7t279fa9ffs26enp3Lt3T9k/KyuLrKwsbt++TWJiovIaUOryYHq+5ORk6taty9WrV/n111+VZWrS09PJyMhQ9s/NzVU+6/fee4+BAwfq5RMfH6/32YuysfCXG/c1PvNcT0pn4YHzjPR0LPbY/O9lYmKi3rW7/+8r/73333+fkSNH0rx5c8zMzPjqq6+4ffs27dq1Y+XKlTRr1owWLVrw0ksvkZiYiKWlJS1btsTV1ZUOHTrwwQcfEBwcjJubG66urtSvX5/ExEQAvXM7OTnRuXNnWrRogb29Pc2aNVO+G3PnzmXs2LFMmTKF7OxsgoKClGHyz7LS+j+p0+lKJN/8HvrE1CyqVzN+aA+9nZ0dERERREdHM378eDw9PWnQoAE7duzQ26+w38LKrqSvpepFJ6wSbqHKzCbHxIhkh9rkpGohtZCuUVFiKsq9a2JiIufOnWPmzJnMnj2bESNGcPfuXeXBUdeuXUlJSSE8PFxvfv6rr75KYmIiFy5cYMGCBUyZMoWQkBDWrFlDr169GDRoEFOnTqVNmzbKdJ+KUN/SUFGuZWVT1P/mkugB9QS69+3btzNgBlj37dv3m40bN75ZAnmLZ0hRwxEfjGzYr18/+vXrp2zfv8TD0aNHgbwbl/vnIdy/duDmzf9bBP3jjz9WXo8dO7ZA8I22bdvStm1bZbuwtUdLkru7O2PGjCE0NJScnByioqJ48803cXJyYseOHXTr1g2dTsdff/1FkyZNsLS0RKst/AbC2dm50IA2Rc3Xgrw5eBMnTmTw4MEsXbqURo0aFVlWf39/Jk6cSM+ePbGwsODmzZsYGxtjYWGhRCsWZScxNavw9LTC0wvj5OSk9725P/jM/e8VNqfU3Nyc9evXF5rv4sWL9bYfbEDke3Du0LBhw5R52vd7/vnnWbduXRG1ECXt/PnzT51HQkomn+2LJyEl/+8xjXO305jk51RkI/TBIF1r1qxBo9Hw22+/0apVK7Kysrh06VKxv1MiT46pqQQcesY5Ojri7u4O5EWzXrFihTLSCYqfn19YTI2kpCSSkpJo06YNAL169WL//v1lXCtRWT11A3Tjxo1jgbEA/+0B/Vgan6KqyP0nAbavQ3dXg4GtHQQNwLCmQ6mdr3nz5gQEBKBWq6lZsyYuLi5YWVmxaNEixo4dy/z585UenyZNmhAUFMSoUaNYvnw5y5Yt08vLzMyMOXPmFAhoU9R8rXz169dn0aJFhISEFLucir+/P8ePH1eWZ6hWrRoLFy7kxRdfxN3dHV9fXzp06CBBiMpI9WrGQFrBdHPjsi9MKVFlZBTsxblvzqqouNadvH1f4zNPQkoW607eLrKHvrAgXSqVik8//ZR79+6Rk5PD4MGDpQEqxCN4MHr1g9vFzc8vLqaGEE9CZqALUYTcfxLQzf0U/kkAQAdwKZbc4ZNLtRE6dOhQRo4cSVpaGj179sTV1bXIHh93d3cOHDigbN/fuwvQrl27AgFtIK+hW1zPctOmTZV8H1yC4f7ekMGDBzN48OAC+T/Y4yVK3wA3e87dTtO7yXewzBvmWBXkr2WoLCehBWNtKhrnetIIrQSepIe+qFExW7ZsKZD24O+UEELf9evXldED27Ztw93dnbNnzyrvJycnK3EyfvnlF65du1ZsfjY2NtjY2HDs2DFat27N1q1bS7sKogop0Qboxo0bDwAHSjJPIcrN9nVK41Px3x5RBo8stdOOHj2ac+fOkZGRQZ8+fWjWrFmpnaskGWbewUITiSo7mRwjK7R2/uSa1CjvYj0zHCxNmOTnlDfHLi2L6uYPn2NXmchahpVbSffQy++NEI/H2dmZ1atXM3LkSBo2bEhwcLDetKOePXsSHByMn5+fMj//YebMmcOIESMwMDBQ1k4X4lEY6HS68jiv7saNG491gEz+rRoq03XM+Wo8xJ4u+EajZqg+Lt25oJXB/dfSMPMOtjdWYJStUd7PNrLjruMguSmsBCrD97LG+UuYFjLfOcPCgjsNXiqHEj2esLCwAksmPWjDhg34+Pjg4PBkIywq8nUsOAc0r4e+uDmgRXkWfm8q8rUUj66iXEdZo/PpVZRrWdn8NwiRwYPpMgRXiCIY2NpR2OMZA1u7Mi9LRWehidS7GQQwytZgoYkk2eH1ciqVqEqehbUMN23aROPGjZ+4AVqRlWQPvfzeCFH+yjpGhqhaqs5/biFKWtAAuBSrPwy3pkNeutCjyk5+rHQhHleyQ22Mtal6w3CzTIxJdqhdbmVKTU0lJCSEmzdvkpuby7Bhw7h06RKRkZGkp6fTqlUrZsyYUSDYx9y5cwvsExERwcmTJwkNDcXMzIwffvgBc/OqtUajg6XJQ5cEehTyeyPE43kwwvnTKq8YGaLqMCzvAghRURnWdMBg+GQMPHygUTMMPHwwkB/XQuUYWT1WuhCPK8fUFI1zPVKr25BhYUFqdZtyD0C0f/9+HBwciIqKIjo6mg4dOvDWW2+xa9cuoqOjSUtLIzIyssBxhe3TtWtX3NzcWLRoEZGRkVWu8VmS5PdGiHJWXIwMIR6B9IAKUQzDmg6lGnCoqtDa+WOcHl9gTpbWzr8cSyWqmoq2lmHjxo2ZPHky06ZNQ61W4+HhQUREBEuWLCEtLY27d+/SqFEjAgIC9I6LiYl56D6iaPJ7I0T50t3VPFa6EA+SBqgQ4qnlmtTgruMgiUopninOzs7s2bOH6OhoZs6ciZeXF6tWrWLXrl3UqVOH2bNnK+vo5UtPT2fcuHHF7iOKJ783QpQviZEhnpY0QIUQJSLXpIYEABHPlISEBGxtbenVqxfW1tasX78eADs7O7RaLREREXTp0kXvmPzGZmH7WFhYkJKSUraVqKTk90aIciQxMsRTkgaoEEII8QTOnj3L1KlTMTAwwNjYmC+++II9e/bg5+dHzZo1cXNzK3CMjY0N/fv3L3Sfvn378sknn1TZIERCiKrBsKYDucMnSxRc8cRkHVBRpuQ6Vh1yLasOuZZVg1zHqkOuZdUg17HqkGv5ZGQdUCGEEKKC0abkEHs6nfS0XMzMDWnUzAwLS1V5F0sIIYQoNdIAFUIIIcqBNiWHIwdSSNXmj0TKIfFONm3aW0ojVAghRJUl64AKIYQQ5SD2dPp9jc88qVodsafTy6lEQgghROmTBqgQQghRDtLTcgtPTy88XQghhKgKpAEqhBBClAMz88L/BZuZyb9mIYQQVZf8lxNCCCHKQaNmZlSz0A8OWM3CgEbNzMqpREIIIUTpkyBEQgghRDmwsFTRpr1lXhTc9FzMzCQKrhBCiKpPGqBCCCFEObGwVNHyVYvyLoYQQghRZmQIrhBCCCGqjPj4eHx9fcu7GEIIIYogDVAhhBBCiGJkZ2cXuy2EEOLRyRBcIYQQQlQp2dnZhIaGcvr0aRo2bMiCBQs4f/48kyZNQqvVYmdnx9y5c6lduza9e/dm4sSJuLm5odFo6NSpE0ePHmXDhg3s3r0brVZLbm4uffv21dteu3YtEyZMIDY2lqysLEaOHElgYGB5V10IISo8aYAKIYQQokq5ePEis2fPxt3dnREjRrBq1Sp2797NypUrqVGjBtu3b2fGjBnMmTOn2HxOnz5NVFQU1atXZ8OGDXrbX3zxBZ6ensyZM4ekpCS6dOlCu3btqFatWhnVUgghKidpgAohKoSYmBjCw8NZs2ZNeRdFCFHJOTo64u7uDkDPnj1ZuHAhsbGxvP766wDk5uZSq1ath+bj7e1N9erVC90+dOgQkZGRhIeHA5CRkcH169dp0KBBSVdHCCGqFGmACiGEEKJKMTDQX1/V0tKShg0bsmPHjgL7qlQqcnNzAUhPT9d778HezPu3dTody5Yto379+iVVbCGEeCZIECIhRIlITU1l4MCBqNVqfH192b59O6dOnaJXr1507NiR/v37c+vWLQDi4uLo168farWawMBALl++rOQxZMgQvL29CQ0NRafTARSZjxClISwsjJ07dz7y/hJ1teK5fv06v/32GwDbtm2jZcuWaDQaJS0rK4vY2FgAnJycOHXqFAARERGPfA4fHx9Wrlyp/E798ccfJVkFIYSosqQBKoQoEfv378fBwYGoqCiio6Pp0KEDEyZMYNmyZezZs4d+/foxY8YMAD788EPeeustoqKi2L59O7Vr1wbybuAmTZrEgQMHuHLlCr/++itZWVlF5iMeXWEPCObOnUvnzp1p0aIFo0ePVm6ke/fuzcmTJwHQaDR4eHgAkJOTw+TJk/H19UWtVrNixQpAHhCIisfZ2ZnVq1fj4+NDUlISgwYNYunSpUyfPh21Wk1AQIDSGB06dChr164lICAAjUbzyOcICwsjKysLtVpNhw4dmDlzZmlVRwghqhQZgiuEKBGNGzdm8uTJTJs2DbVajY2NTaFzrlJSUrh58yadOnUCwMzMTMmjefPmODo6AtCkSRPi4+OxtrZ+orlbQl/+A4K1a9cCcO/ePby9vRk+fDj29va88cYbREZGEhAQUGQe33zzDfHx8fz4448YGRmRmJioPCB43OAuZS01NZWQkBBu3rxJbm4uw4YN49KlS0RGRpKenk6rVq2YMWNGgaGbc+fOLXSfU6dOMWLECCCvJyxfTk4O06dP5/Dhw2RmZhIcHMzAgQPLtK7POicnJw4dOlQgvWnTpmzZsqVAev369YmKilK2x4wZA0C/fv3o16+fkv7gtrm5uTQ6hRDiCUgDVAhRIpydndmzZw/R0dHMnDkTT0/PQudcpaSkFJmHiYmJ8lqlUpGdnY1Opyty7pZ4dA8+IPDw8CAiIoIlS5aQmZnJnTt3aNSoUbEN0J9//pmBAwdiZJT3r6N69eqcPXu2UjwgKK4BDnm98oU1wN96661C9xkxYgRTp06lTZs2TJkyRdl//fr1WFlZsWvXLjIyMnjttdfw8fHh+eefL6OairKQ+08CbF+H7q4GA1s7CBqAYU2H8i6WEEJUCtIAFUKUiISEBGxtbenVqxfW1tasWbNGmXPVqlUrsrKyuHTpEo0aNeK5555jz549dOzYkYyMDCUASGGcnZ2LzEc8Ont7e9566y1q1arFzJkz8fLyYtWqVezatQs3NzfGjh1LRkYGUHxQlgfd/4AgJiYGY2NjJfpoRVJcAzwtLY27d+8W2gCPiYkpsI+HhwdJSUm0adMGgF69erF//34ADh48yJkzZ5S5hMnJycTFxUkDtArJ/ScB3dxP4Z8EAHQAl2LJHT5ZGqFCCPEIpAEqRCUXHx/Pb7/9Ro8ePcq1HGfPnmXq1KkYGBhgbGzMF198gUql4tNPP+XevXvk5OQwePBgGjVqxIIFCxgzZgxfffUVRkZGLF26tMh8TUxMWLp0aaH5iEd38eJFNmzYwIEDB7C2tmb9+vUA2NnZkZKSQkREBF26dAH+F5SlRYsWekFZ2rVrx9q1a2nbtq0yBPf+BwSHDx/GzMwMa2vrCnd9Huyhv78BXqdOHWbPnq00wPOlp6czbty4YvcpzNSpU2nfvn0p1USUu+3rlMan4r89ogweWWqnjY+P580336R169b89ttvODg4sGLFCgYOHMjEiRNxc3NDo9HQqVMnjh49SmxsLCNGjCAzM1OJ2Ovg4FBgKHpQUFCplVkIIQojDVAhKons7Gxl6OP94uPj2bp1a7k3QNu3b1/oTXdhc65eeuklNm3apJf2wgsv0LZtW2V72rRpyuui5m6JRzdt2jQuXLjASy+9hLW1Nd7e3uTm5uLi4sKLL75IixYtiI+Pp1evXqhUKjZv3sycOXNo2bIlt27dws/Pj2XLlnHp0iXatGlDYmIi1atXx9nZmS+//JLPPvuM06dPA7By5UoWL15M/fr1+eSTT7h+/ToAkyZNKrfe0Qd76O9vgGu1Wr0GeL78xuaD+9jY2GBjY8OxY8do3bo1W7duVY7x8fFhzZo1eHp6YmxszMWLF3nuuecKLOchKi/d3cIDFRWVXpLi4uJYvHgxs2bNIiQkhF27dhW579q1a3nnnXfo2bMnmZmZ5OTkEB0dXWAouhBClDVpgApRBpYsWYKJiQnvvPMOn332GX/99RebNm3i559/5rvvvmPv3r2cP38egJ07dxIVFcW8efMICwvD1NSUP//8k1atWhEYGMinn34K5K1zt2XLFqZPn86FCxfw9/enT58+vPvuu+VZ1RKVkvE3p//ZTFrWXcyNbWlWszeWphVvfmFlMG/ePIKDg4mOjiYiIoK1a9dy4sQJNBoN3bp145NPPuHixYu88847HDhwAFtbW9q2bUuTJk1YuXIlX3/9NatXr2by5MmEhYVhY2ODgYEB3377rZLn7NmzsbCwYOjQoQB88MEHDBkyhNatW3P9+nX69+/PwYMHy6X+hfXQ79mzBz8/P2rWrImbm1uBY2xsbOjfv3+h+8yZM4cRI0ZgYGCgF4Sof//+xMfH07FjR3Q6HXZ2dkq0YFE1GNja5Q27LSS9tDk5OdG0aVMAXF1diY+PL3LfV155hQULFihB31566aVCh6ILIURZkwaoEGWgdevWLF26lHfeeYdTp06RmZlJVlYWx44dw8PDg7179xZ57M2bN9m+fTsqlYrg4GCmT5+Ou7s7Wq0WU1NTxo0bR3h4OGvWrCnDGpW+lIy/OXBlBtqsv5W0O6kXaf/CGGmEPqVjx47x2muvoVKpqFmzJu3atePkyZNYWlri5uamLIvzwgsvKI2rxo0bExMTA+T9Tb733nv8/fffZKRnUt22DjHRydyMz+T5eubKeX766SfOnTunbKekpKDVarGwsCjD2uYprIfezc1NiXh6v3nz5imvx4wZU+g+rq6uepFTJ0yYAIChoSFjx45l7NixJVRyUeEEDYBLsfrDcGs65KWXMlNTU+W1SqUiPT29yDnbPXr0oEWLFuzbt4+BAwcyY8YMvLy8CgxFzw+yJYQQZUUaoEKUAVdXV06fPk1ycjImJiY0a9aMkydPcvToUb0ImoXp2rUrKpUKAHd3dyZNmkSPHj3o1KmTsmRJVXT6n816jU8AbVZej+irdd8vp1JVffdHIjY0NFS2DQ0Nyc7OBmDixIm8++67eLb1Y+XSaDZvX8ydf3K4l5TLlYsZaFNysLDMuynesWOH3lI7VV1SUhJHjhwhJSUFS0tL2rRpg42NTXkXS5Qgw5oO5A6fXGGi4BY1Z/vKlSu88MILvPPOO1y/fp0zZ85Qv379QoeiCyFEWTIs7wII8SwwNjbGycmJjRs30qpVK1q3bk1MTAyXL1+mQYMGemsPPhjk5P65Y6GhocyaNYv09HRee+01Lly4UGZ1KGtpWXcLTU8vIl0Uz8LCQlkCx8PDgx9++IGcnBzu3LnDzz//TPPmzR85r3v37uHg4EDs6XQO/LRNSTczsyA5WUvs6bxeGB8fH1auXKm8/8cff5RQbSqGjz/+WK+HNykpia1btxIbG8v169eJjY1l69atJCUllWMpRWkwrOmA4eCRqD6ehuHgkeUa/Xbo0KGsXbuWgIAANJr/zUPdsWMHvr6++Pv7ExsbS+/evTl79ixdu3bF39+fuXPnMmzYsHIrtxDi2SU9oEKUEQ8PD8LDw5k9ezYuLi5MmjQJV1dXDAwMqFmzJufPn1cidRY1RPHy5cu4uLjg4uLCiRMnuHDhAo6Ojmi12jKuTekzN7YtNN2siHRRPDs7O9zd3fH19aVDhw64uLjg7++PgYEB06dPp1atWo/8QGPkyJGEhISgMrSk/ose3NbkBRlyfbk9y1YPJ3TEfubOm86UKVMYN24carWWFrx0AAAgAElEQVSa7OxsPDw8mDFjRmlWs0x99dVXettHjhwpENTl3r17HDlyhMDAwLIsmqiCnJyciI6OVrbz51oDesPB84eMh4aGEhoaqpdHUcHihBCiLEkDVIgy0rp1axYsWECrVq2oVq0apqamtG7dGoCxY8cSHByMnZ0dbm5uRTYov/76a2JiYjA0NKRhw4Z06NABQ0NDDA0NUavV9O3bt8oEIWpWszd3Ui/qDcO1MK5Fs5q9y7FUldvixYv1tidOnAjkrRF6+/Zt2rZtqxeJePPmzcrr+98LDAwkMDCQ/xzWcv1qlrJP7VovMnHUVuo8b0xLj7yHKOHh4aVWn7ISHx/PgAEDlKH0DRs2ZMGCBcryF7du3WLWrFkkJSWRlpZGTk4OQUFB7NmzB8ibqzdo0CCuX7/OunXrWLduHZmZmdSrV48FCxZgbm7+kBII8XQkoJsQoiIx0OkKi+VW6nQ3btx4rAPyb5BE5SbXsWRUhJuJsriW+fVMz7qLmdw0lZonvZbalByOHEghVfu//yPVLAxo094ybw7of9dGrAjz5J5GfHw8bdq0Ydu2bbi7uzNixAgaNmxIVFSUsv4iwN69e5kxYwbPP/88LVu2VI4/efIkL774IhMnTkSj0WBnlxctdcaMGdSsWZNBgwaVSDnl97XqKMlrWVhANwvjWhLQrQzId7LqkGv5ZP4bq8TgwXTpARWiknmWosNamtaSgEMVmIWlijbtLYk9nU56ei5mZoY0amamND51cz9VIoXqAC7Fkjt8cqVshDo6OiprmPbs2bPQpVX+/PNPqlWrptf4vHz5MomJiSxbtgyA2NhYZs6cyb1799BqtXpLuAhRGiSgmxCiopEGqBCVjNxMiIrEwlJFy1cLmbO8fZ3+MhWQt719HQweWTaFK0H3BworbPvQoUNERUWxcuVKTpw4gVarJTExkV9//ZVt27YpkayHDx/O8uXLadKkCRs2bODw4cNlVgfxbJKAbkKIikai4ApRycjNhKgMdHc1j5Ve0V2/fp3ffvsNQBmKm+/atWuMHz+epUuXUrt2bQIDA/Hz8+Pbb79l4cKF1KhRQ9k3JSWF2rVrk5WVxdatW8u8HuLZIwHdhBAVjfSAClHJyM2EqKgaNGjA+fPnATCwtaOwCAMGtnZlW6gS4uzszOrVqxk5ciQNGzYkODhYiTy6ceNGEhMTlbmcDg4OdOvWjWvXrjFq1Cglj8jISEaNGkXXrl2pUaMGLVq0UJbGEaK0SEA3IURFIw1QISoZuZkQlULQALgUqz8Mt6ZDXnolZGRkxMKFC/XS8qMEu7m5MWLEiALH9O3bt0BacHAwwcHBpVNIIQphaZoXcEgCugkhKgppgApRycjNhKgMlm7eyg+/xpF5528CX6zDx539IGgA879Zz5YtW6hRowaOjo64uroydOhQevfurUSU1Wg0dOrUiaNHj7Jhwwb27t1LamoqcXFxDB06lMzMTL7//ntMTExYu3Yt1atX5/Lly4wfP547d+5gbm7OrFmzqF+/fnl/DHqqSlRgUflIQDchREUiDVAhKiG5mRAV2cGDB4mLi2PXjz+i0+l46623ONbUE7PrCezatYvIyEiys7MJDAzE1dX1ofnFxsayd+9eMjIy8PT0ZNy4cfz444989tlnbN68mSFDhjB69Gi+/PJLXnrpJf7zn/8wduxYNm3aVCL1cXJyIjo6+qnyqGpRgYUQQognJQ1QIYQQJergwYMcPHiQgIAAAKX3MiUlhcDAQMzMzADw9/d/pPzatm2LpaUllpaWWFlZKce5uLjw119/odVqOX78OCEhIcoxmZmZJVyrp1TFogILIYQQT0oaoEIIIUqUTqcjNDSUgQMH6qX/+9//LvIYlUpFbm4uAOnp6XrvmZiYKK8NDQ0xNTUF8pZCycnJITc3F2trayIjI0uqCiWuqkUFFkIIIZ6ULMMihBCiRLVv354NGzag1WoBuHnzJrdv38bd3Z3IyEjS09PRarVKFFnIG+Z66tQpACIiIh7rfFZWVjg5ObFjxw4grwH8559/llBtSkZR0X8ra1RgAbNmzeLQoUPlXQwhhKh0pAdUCCFEifLx8eH8+fN0794dgGrVqrFw4UKaN29OQEAAarWamjVr4uLigpWVFQBDhw5l6NChrFu3Dj8/v8c+56JFixg7dizz588nOzuboKAgmjRpUqL1eipVLCqwQG+JHSGEEI/OQKcrbKW2Uqe7cePGYx1gb2/P7du3S6k4oqzIdaw65FpWHWV5LbVaLRYWFqSlpdGzZ09mzpxJs2bNyuTc5a20o+DKd/LpxMfH8+abb9K6dWt+++03HBwcWLFiBRcvXuSTTz4hPT2dF154gdmzZ2Nra0tYWBhqtRpvb2+6dOnCypUrqV+/Pu+//z6enp4MGPDkDxfkWlYNch2rDrmWT8bR0RHA4MF0GYIrhBCizIwePRp/f38CAwPp3LnzUzc+DTPvYJXwHbbX/o1VwncYZt4poZKWPMOaDhgOHonq42kYDh4p0W8roLi4OIKDg9m/fz/W1tbs2rWLsLAwxo8fT1RUFI0bN2bOnDl6x1hbWzNt2jSGDx/O9u3bSUpKeqrGpxBCVHXSABVCCFFmFi9eTGRkJIcOHeLDDz98qrwMM+9ge2MF5iknMUm/hHnKSWxvrKjQjVBRsTk5OdG0aVMAXF1duXLlCklJSbz66qsA9OnTh6NHjxY4ztvbGxcXF8aNG8esWbPKtMyVyfLly/Hx8SE0NFQv/eTJk0ycOPGx8+vduzcnT54sqeIJIcqIzAEVQghRKVloIjHK1o8ia5StwUITSbLD6+VUKlGZ5UdYhrzIzElJSY90XG5uLufPn8fc3JykpKT8YWfiAatXr+a7777T+3yys7Nxc3PDzc2tHEsmhChL0gMqhBCiUlJlJz9WuhCPy9raGhsbG6XX8/vvv6dNmzYF9lu2bBkNGjRg8eLFjBgxgqysrLIuaoU3ZswYrl69ysCBA2ncuDEffvghQUFBfPTRR8TExPB///d/ABw+fBh/f3/8/f0JCAggJSUFyBs94efnh1qtZvr06Uq+O3fupEuXLnh5eRXaOy2EqHikB1QIIUpAfkCSrl27lsn5EhISmDhxYrFra1Z1OUZWj5UuxJOYN2+eEoTo+eefLzAH9MKFC6xfv56IiAgsLS3x8PBg/vz5fPzxx+VU4oppxowZHDhwgE2bNrFy5UoiIyPZunUr5ubmxMTEKPuFh4czffp03N3d0Wq1mJqaEh0dzd69e9m5cyfm5uYkJiYq+2dnZxMREcG+ffuYM2cOGzZsKI/qCSEegzRAhRCiEnJwcHimG58AWjt/jNPj9YbhZhvZobXzL8dSicrKycmJ6OhoZXvo0KHK6507dxbYf968ecrrgwcPKq8///zz0ilgFRMQEIC5uXmBdHd3dyZNmkSPHj3o1KkTjo6O/PTTT/Tr10/Zv3r16sr+nTt3BvLm7F67dq1sCi+EeCoyBFcIIZ7Apk2bUKvVqNVqJZjO0aNH6d69O6+++qpyw6rVaunbty+BgYH4+fmxd+9eIG/JB29vb8LCwvDy8iI0NJRDhw4RFBSEp6cnv//+OwCzZ8/mww8/pFu3bnh6erJu3TrleF9fXwDS0tIICQmhffv2vPPOO3Tt2lUJzNGgQQO+/PJLpXf2n3/+UY7v06cParWawMBArl+/DuT15E6cOLFAPW7dukXPnj3x9/fH19dXb6hbamoqAwcORK1W4+vry/bt2/npp58ICAjAz8+PESNGkJGRAYCHhwcaTV6D8eTJk/Tu3bvYegIsWbKEzp07o1ar+eqrr5T0XJMa3HUcRJqlG5lmL5Fm6cZdx0HkmtQomYssxEOoMjKwvXKVGucvYXvlKqr//p2Lh6tWrVqh6aGhocyaNYv09HRee+01Lly4UGw+JiYmQN6c3ezs7BIvpxCi5EkDVAghHlNsbCzz589n48aNREVFMXnyZCCvkbZt2zZWr17NF198AeQFNVm+fDl79+5l06ZNTJ48mfz1ly9fvkxISAiHDh3iwoULbNu2jW3btvHpp5+ycOFC5Xxnzpxh48aN7Nixg7lz55KQkKBXntWrV2NjY8OBAwcYNWoUp06dUt5LTU2lZcuWREVF0aZNG6VhN2HCBPr06UNUVBRvvPGGXgTKwuqxdetWfHx8iIyMJDIykiZNmij779+/HwcHB6KiooiOjqZDhw4MHz6cJUuWsG/fPrKzs1mzZs1DP9fC6nnw4EHi4uKIiIjgxx9/5NSpUxw5ckQ5JtekBskOr3O37hCSHV6XxqcoM6qMDOwuxlEtMQlTrZZqiUnYXYyTRuhTunz5Mi4uLnzwwQe4ublx4cIFvL292bBhA2lpaQB6Q3CFEJWPNECFEJXKxx9/zLlz58q1DL/88gtdu3bFzs4O+N9wsI4dO2JoaEjDhg2VnkadTqf0QPbr14+EhATlPScnJ1xcXJRjvLy8MDAwoHHjxsTHxyvnCwwMxNzcHDs7O9q2bcuJEyf0ynPs2DGCgoIAaNy4MS4uLsp7JiYm+PvnDUlt1qyZMkTt+PHj9OjRA4ABAwZw7Ngx5ZjC6tG8eXM2btzI7NmzOXPmDJaWlsr+jRs35tChQ0ybNo2jR48SHx/P888/j7OzM1D00hUPKqyeBw8e5ODBgwQEBBAYGMjFixeJi4t7aF5ClDarhFsYZ+oHGzLOzMIq4VY5lahq+Prrr/H19UWtVmNsbEyHDh3o0KEDAQEBdOrUCX9/f8LDw8u7mEKIpyBzQIUQlcr9QzArmvyhYIDSy7llyxbu3LnD7t27MTY2xsPDQxmOev+SD4aGhsrxhoaG5OTkKO8ZGBjonefB7eIYGRkp+z/qELXC6tGmTRu+//579u3bx/Dhw3n33Xfp06cPAM7OzuzZs4fo6GhmzpyJp6dnseXJzc0FUD6HouplYGCATqcjNDSUgQMHPkJthSg7qszCv0tFpQuUB1EjR47US2/bti1t27YFYOrUqYUeGxoaWmD90M2bNyuv7ezsJAquEJWE9IAKIUpcYXMC586dS+fOnfH19WX06NFKw+b+hcQ1Gg0eHh4A5OTkMHnyZOVJ+IoVK/T2z8nJISwsDF9fX/z8/Fi2bBmQt9B5+/btUavVvPfee0p5RowYQZcuXQgICFDmYT4pT09Pdu7cqcxlLG44WHJyMvb29hgbG/PLL788UZCMvXv3kp6ejkaj4fDhwwXWy3N3d2fHjh0AnDt3jrNnzz40z1atWrF9+3YA1q9fr3zuRbl27Ro1a9ZkwIAB9O/fn9OnTyvvJSQkYG5uTq9evRg6dCjHjx8nPj5e6am8f+mKunXrKkOEIyIiHlrP9u3bs2HDBrRaLQA3b97k9u3bD62fEKUtx6TwZ/hFpYuSJ3Nwhaic5FdSCFHi8ucErl27FoB79+7h7e3N8OHDAfjwww+JjIwkICCgyDy++eYb4uPj+fHHHzEyMirQyPvzzz9JSEhQolbmLxi/ePFiDh8+jKmpqZI2f/58PD09mTNnDklJSXTp0oV27doVGQTjYRo1asRHH31E7969MTQ0pGnTpkXu27NnT4KDg/Hz88PV1ZX69es/9vlcXFzo06cPGo2GsLAwHBwc9IboBgcHM2zYMNq3b0/9+vVp2LAhVlbFL0UydepUhg8fTnh4OLVr12bmzJnF7h8TE0N4eDhGRkZYWFgwf/585b2zZ88ydepUDAwMMDY25osvvuDevXuEhISQk5ODm5ub0oM5YsQIRo4cyaxZs3j11VcfWk8HBwfOnz9P9+7dgbzAJQsXLsTe3v6xPkMhSlqyQ22Mtal6w3CzTIxJdqhdjqV6duTPwVU+fy0Ya1PRONcj577RJUKIiscgvxeijOlu3LjxWAfY29vLU+8qQK5j1VHctbx48SL9+/ene/fuqNVqPDw8iIiIYMmSJaSlpXH37l3efvttQkND6d27NxMnTsTNzQ2NRkOnTp04evQoQ4YMYeDAgXh7e+vlnb//Cy+8oPSo+vn54ePjg6GhIQMGDKBatWp07NiRjh07YmFhQadOncjIyEClUgFw9+5dvv32Wxo0aFDqn9PTmj17NhYWFnpLQjwoJyeHrKwszMzMuHz5Mq+//jqHDh3SG0pbnIrwvSysnoaZd7DQRKLKTibHyAqtnb8EGSpGRbiOzxpVRgZWCbdQZWaTY2JEskPtEmn8yLV8ONsrV6mWmFQgPbW6DXdfeL4cSlSQXMeqQ67lk3F0dAQoMG9IekCFECXuwTmBXl5erFq1il27dlGnTh1mz56tzP9TqVTKnMD09PRHPoetrS2RkZEcOHCAtWvXsmPHDubMmcOaNWs4cuQIkZGRLFiwgH379qHT6Vi2bNkT9T5WBmlpafTp04esrLyegOnTpz9y4/NhkpKSOHLkCCkpKVhaWtKmTRtsbGxKJO/iGGbewfbGCr01Po3T42WZFVGh5JiaVpjGzrNG5uAKUXlJA1QIUeISEhKwtbWlV69eWFtbs379eiAvSIRWqyUiIoIuXboAeZFgT506RYsWLfTmBLZr1461a9fStm1bZQju/YuPazQajI2N6dKlC87Oznz44Yfk5uZy48YNPD09ad26NT/88ANarRYfHx9WrlypDBP9448/ih02W5E8GKyjMJaWluzevbvEz52UlMTWrVu5d++eknbz5k169OhR4o3QB+tpoYnUa3wCGGVrsNBEkuzweomeWwhR+eSYGIG2iHQhRIUm31IhRIkrbE7gnj178PPzo2bNmnpBdIYOHcrQoUNZt24dfn5+Snr//v25dOkSarUaIyMjBgwYwNtvv628f/PmTUaMGKH0no4dO5acnBw+/PBDkpOT0el0DBo0CBsbG8LCwvjss89Qq9Xk5ubi5OT0SOtSPuuOHDmi1/iEvPm8R44cITAwsFTPrcpOfqx0IcSzRebgClF5yRxQUabkOlYdci2rjqKu5ffff8/169cLpNetW5eePXuWapmsEr7DPOVkgfQ0SzfpAS2CfCerDrmWj6a05uCWFLmOVYdcyydT1BxQWYZFCFGllWWY/vxIrU9j9uzZFWaRdUtLy0LTLSwsSv3cWjt/so3s9NKyjezQ2vmX+rmFEJVD/hzcOw1e4u4Lz1eoxqcQomgyBFcIUWWVdZj+H374ocTzLE9t2rTh5s2besNwra2tlTU9S1OuSQ3uOg6SKLhCCCFEFSM9oKLExcfH4+vr+8j7b9iwgfHjx5diicSzyirhlt78IADjzCysEm6VyvkaNGiAVqulb9++BAYG4ufnx969e4G874W3tzdhYWF4eXkRGhrKoUOHCAoKwtPTk99//13J588//6Rbt254enqybt06AHQ6HVOmTFGWndm+fTuQtz5n7969GTJkCN7e3oSGhlJSUytsbGzo0aMHjRo1om7dujRq1KhUAhAVJdekBskOr3O37hCSHV6XxmclkZSUxKpVqx77uAULFuhtl8SIAiGEEBWP9IAKIaqs8gjTb2pqyvLly7GyskKj0dCtWzcCAgIAuHz5MkuXLmXOnDl07tyZbdu2sW3bNn788UcWLlzIihUrADhz5gw7duwgLS2NgIAA/Pz8OH78OH/++SeRkZFoNBo6d+6s9ET+8ccfREdH4+DgQFBQEL/++iutW7cukfrY2NiUesAhUbXcu3ePNWvW8NZbbz3S/jqdDp1Ox8KFC/noo4+U9JIaUZCdnY2RkdzuCCFERSE9oKJU5OTkMGrUKDp06MAbb7xBWloavXv35vjx40DeEhoeHh7K/rdu3WLAgAF4enoydepUJb1BgwZ8+eWXqNVqunbtyj///FPmdRGVV1Hh+EszTL9Op1P+Zvv160dCQoLyd+vk5ISLiwuGhoY0bNgQLy8vDAwMaNy4MfHx8UoegYGBmJubY2dnR9u2bTlx4gTHjh3jtddeQ6VSUbNmTdq0acPJk3lBepo3b46joyOGhoY0adJELy8hytr06dO5cuUK/v7+fPbZZ0WOCGjXrh0fffQRvr6+jBw5kvT0dPz9/QkNDQXyfv+h+F7+U6dO0atXLzp27Ej//v25dStvdEPv3r359NNP6dSpE19//XU5fApCVB3Lly/Hx8dH+W4+DQ8PDzQazcN3JO93YuvWrU99zgfJyLvyJ48ERamIi4tj8eLFzJo1i5CQEHbt2lXs/n/++Sd79+7FxMQEb29v3n77berUqUNqaiotW7bkk08+YerUqaxbt46wsLAyqoWo7MojTP+WLVu4c+cOu3fvxtjYGA8PDzL+G/jI9L55p4aGhpiYmCivc3JylPcMDPQDxj24/aD8fABUKhXZ2bIQuyg/48aNIzY2lsjISLKzs0lLSyt0REBcXBzz5s3jlVdeAWDnzp1ERkYWmmdhvfwtWrRgwoQJrFy5kho1arB9+3ZmzJjBnDlzAMjKyiqV9XGFeNasXr2a7777Lj+iKVA2IwvyG6A9evQo1fOIsic9oKJUODk50bRpUwBcXV0f2iPj5eWFtbU1ZmZmNGzYUFn6wcTEBH//vKiXzZo149q1a6VbcFGl5JiaonGuR2p1GzIsLEitblNqAYjyJScnY29vj7GxMb/88ssT/c3u3buX9PR0NBoNhw8fxs3NDQ8PD3744QdycnK4c+cOR48epXnz5qVQAyFKTnEjAurWras0Ph+msF7+ixcvEhsby+uvv46/vz8LFizg5s2byjEyh1SIpzdmzBiuXr3KwIEDady4MR9++CFBQUF89NFHBXoS/+///o+YmBgADh48SLdu3QgMDOTdd99Fq9Uq+61YsUIZFXHhwgUADh8+jL+/P/7+/gQEBJCSksL06dM5duwY/v7+LFu27InOd+LECbp3745araZLly6kpKTo1S8qKopu3bo9cq+sKBnSAypKxf09PSqVivT0dFQqFbm5uQCkp6fr7X9/D46hoaHSg2NkZKT0/kjPjngS+WH6y4KBgQE9e/YkODgYPz8/XF1dqV+//mPn4+LiQp8+fdBoNISFheHg4ECnTp04fvw4/v7+GBgYMH78eGrVqqX88xaiIipuREC1atUeOZ/Cevl1Oh0NGzZkx44dhR7zOPkLIQo3Y8YMDhw4wKZNm1i5ciWRkZFs3boVc3NzNmzYUOgxGo2G+fPns2HDBqpVq8bixYtZtmwZw4cPB8DOzo69e/eyatUqwsPD+eqrrwgPD2f69Om4u7uj1WoxNTVl3LhxhIeHs2bNGoDHPt8HH3zAe++9x5IlS2jevDnJycmYmZkpx+3evZtly5axdu1abG1tS/iTE8WRBqgoM05OTvznP/+hXr16RERElHdxhChRGo0GW1tb7Ozsirwhjo6OVl7PmzdPee3k5KS8N3LkyEKPNTAwYOLEiUycOFEvvW3btrRt21bZnjZt2hPXQYiSYGFhofQyPM6IAGNjY7KysjA2Nn6k8zg7O6PRaPjtt99o1aoVWVlZXLp0iUaNGpVIPYQQBQUEBGBubl7sPsePH+fcuXMEBQUBecPh7x/t0KlTJyBvhFz+MHl3d3cmTZpEjx496NSpk95w34cp6nwXL16kVq1aymghKysr5ZhffvmFkydPsn79er10UTakASrKzNChQ/nggw9YunQpfn5+5V0cIUpMQkICvXv3ZujQoWV/7pRM1p28TWJqFtWrGTPAzR4HS5OHHyhEKbGzs8Pd3R1fX1/c3Ny4cOHCI40IGDBgAGq1mmbNmrFo0aKHnsfExISlS5fy6aefcu/ePXJychg8eLA0QIUoRfePLDAyMlJGtgHK6AadToe3tzf/+te/Cs0jf5ScSqVS4h+Ehobi5+dHdHQ0r732Gt9++22B4x73fGfOnCmyHi+88AJXr17l0qVLuLm5FVtnUfKkASpK3P29OYDeTfnx48e5ffs2kDevAKBfv37069dP2Sd/qAXA+fPnldddu3ala9eupVZuIZ6Ug4MDP//8c5mfNyElk8/2xZOQkh9kKY1zt9OY5OckjVBRrhYvXqy3vXz5ctasWUOzZs1wcnIC9EcEAIwfP15vflf+739xvfxNmzZly5YtBc6/efPmp6+EEKJYTk5OrF69mtzcXG7evMmJEycAeOWVVxg/fjxxcXHUq1eP1NRUbt68ibOzc5F5Xb58GRcXF1xcXDhx4gQXLlzA0dFRb+7ok5zv77//5sSJEzRv3pyUlBRlCG7dunWZOHEigwcPZunSpfLgqoxJA1RUWIaZd7DQRKLKTibHyAqtnb8sRC/EfdadvH1f4zNPQkoW607eZqTnow9fEqK0lUUUTRkNIETZcnd35/nnn6d9+/Y0aNCAZs2aAVCjRg3mzp3LBx98QGZmJgCjR48utgH69ddfExMToyxT1qFDBwwNDTE0NEStVtO3b1+GDBny2OdbsmQJEyZMID09HTMzM715pPXr12fRokWEhISwatUqXnzxxVL6pMSDDPLX0ipjuhs3bjzWAfb29krPmai8HvU6GmbewfbGCoyy/xeVLNvIjruOg6QRWkHId7L8TYi8wum/0wqkN6ttzlT1C4+cj1zLqqGiXscxY8awYcMGnJ2duX79Ov7+/ly9epU6deowduxYRowYQWJiInZ2dsydO5c6deoQFhaGmZkZf/zxB3fu3GH27Nls3ryZ48eP06JFC2UO9SeffMLJkydJSU1D1dATm/YDAbh75ig3d4XjZG9NW4/WXLlyhTVr1pCamsqECROIjY0lKyuLkSNHEhgYWJ4fT6Eq6rUUj0euY9Uh1/LJ/PehY4G15GQZFlEhWWgi9RqfAEbZGiw0ha8RJ4pWGgsuz5o1i3379pVonuLxVa9WeLCW6uaPFsRFiHyzZs3i0KFDpZL3jBkzqF27Nps2bWLIkCGcP3+e7777jn/9619MmDCBPn36EBUVRc+ePfWCbCUlJbFjxw4+//xz3n77bYYMGcL+/fs5e/Ysf/zxB5DXuN29ezfdp67l5tnfSb15kdysTK5smctLb0+n82cruXPnjpLn/Pnz8SqAHscAACAASURBVPT0JCIigk2bNjFlyhRSU1NLpd5CiIpHlZGB7ZWr1Dh/CdsrV1H9dx6pKFsyBFdUSKrs5MdKF2Vr1KhR8jSwAhjgZs+522l6w3AdLPOGHgrxOEaNGlVm57o/iubx48f5+uuvAejVq9f/s3fnAVFW++PH32wCgqC4RCp6E9dUyBULBJUBNHcTvElkmetNE8SruaQ/9zQRNUu01NxS3FJwH3fNpdSuYvXFDRVZ3ECEAWSZ+f2BTCCDgrL7ef3jzJnznDkPjwPzec45n8OsWbO09bK3HGratCk1atSgWbNmADRu3Jg7d+7QokULQkND2bBhA1GPUkiNf0Dq3Vug0WBs9SbGVm8Sn5JOnz59WL9+PQDHjx9HqVQSFBQEZCUxiYqKolGjRiV2/kKI0mHw5AlW1yMwSnv6N1MFRqrkYt8fXOQlI6CiTMo01J0SO7/y11FycjI+Pj4oFAq6dOnCzp07891w+e7du3h7e+Po6Kj9gpc9sgBZay/effddAG7duqVNZX7p0iU++OADunbtysCBA7l79y4Avr6+2sQfDg4OLFiwIM+m0qL4WZtXYrqrDc7/sqDlG6Y4/8tCEhAJACIjI3F2dmbUqFG4uLgwdOhQUlJSCAwM5P3336djx468/fbbZC/D8fX1ZdeuXUDBPtNLlix56b4VdH/O7L0/9fX1c+0tnb1X9O3bt1m+fDnBwcF8unATlk0dUGek5Wrj2dkAGo2GFStWoFQqUSqV/P777xJ8CvGaqBJ795/g8ymjtHSqxN4tpR69viQAFWWSysqNDEOrXGUZhlaorNxKqUdlz5EjR7C2tubgwYMcPnyYzp07M3LkSGbMmMHBgwfZtGmTNtvbn3/+ybJlyzh06BAhISFERUXh4ODAb7/9BsDZs2epVq0aMTExnD17lg4dOpCens6UKVNYsWIF+/btY8CAAcybN09nX7I3lfbx8dGOLIiSYW1eCX/H2sxS1MffsbYEn0Lr+vXrDBo0iGPHjlGlShXWrFnDJ598wp49e/j555/RaDQolbqXNVhZWbFnz558P9PffvttkfSxbdu27Ny5E4Dt27fj4OBQ4GMTExMxNTXFwsKCrnU0JF75HQCTmjY8iYvB4skDvO1rEBISoj3GxcWF1atXawPv7Km8QoiKzyAto1DlovjIFFxRJqkrVedR7cGSBfc5mjZtyowZM5g9ezYKhQILC4t8N1x2cnLCwsICyJq+FhUVRfv27VGpVCQlJRETE0OfPn04e/Ysv/32G926deP69euEh4fz73//GwC1Wk2tWrV09kXXptJCiNJVu3Zt2rVrB0C/fv1YtWoVNjY2LFu2jMePH6NSqfjmm2+YPXs2KSkpODs74+DgQEJCAuvXr6d69erEx8ezc+dOLly4gKurK5MnT2bOnDmkpqbi5uZGkyZNCrRnZ35mzZqFn58fQUFB2iREBdW8eXNatGiBs7MztWvXxrFDeyrVMKV23arYDP+S/62ayCebzXLt8efr68u0adNQKBSo1WpsbGxybf0lhKi4MisZgiqfclGi5Ccuyix1peokWv+7tLtRZtna2rJv3z4OHz7M/PnzcXR0zLdu9lQ2+Gf6GmSNPgQHB9OgQQMcHBzYtGkT58+fZ+rUqURFRdG4cWNCQ0Nf2Bddm0qXZ5mZmRgYGJR2N4R4JXp6enmeT5o0iT179qBWq+nQoQMtW7Zk4cKFODo6cvToUSDrd8TWrVtJS0tj6tSpvP3222zbto0PP/yQffv2MWnSJFavXp3v6KkuZ8+eBcDf3z9Xed26ddmyZUue+tlZbiHv3tI5X8v5OCfVuzUwG/NvNBoNkyZNws7ODgBTU1Pmz59f4H4LISqOROs3MFIl55qGm17JiETrN0qxV68nmYIrRDkVGxuLqakpH3zwASNGjOCPP/7QbrgMkJSUpA0089O+fXuCgoLo0KEDLVq04NSpU1SqVAkLCwtsbW2Ji4vj3LlzAKSnpxMeHl7s51VU8lsDd+LECdzd3XF1dWXs2LE8eZoBz8HBgdmzZ+Ph4cGuXbs4cuQIHh4e2v3HhChvoqKitJ/fHTt2aEdDraysSE5OxtDQkDp16gBQv3597VrP7DWaFy9exM7ODiMjIwwNDenXrx9nzpwphTMpvA0bNuDm5kbnzp1JTEzEx8cH9f1Y1D8GkLlgMuofA1Dfjy3tbgohSlCmsTFxtm+RXM2SJ2ZmJFezlAREpURGQIUop/7v//6PWbNmoaenh5GREXPnzkWj0eS74bIuDg4OREdH4+DggIGBAbVr16Zhw4ZA1qjp8uXLmTp1Ko8fPyYzM5MhQ4bQpEmTkji9InH9+nUCAgJo164dY8eOZfny5axfv167J+EXX3zB2rVrGTp0KADVqlVj//79PHz4EA8PD7Zv3069evWIj48v5TMRovBsbW1Zs2YN/v7+NG7cmEGDBpGQkICrqysWFha5ZkbAPyOmz46clkfDhg1j2LBh2ufq+7FoAqfC06BTA3AjHLXfDPRrWpdOJ4UQJS7T2JhH9euVdjdee3rZC/FLmCY6OrpQB8iWDxWDXMeKo6xfy8jISPr168fvv2clJjl58iSLFi1CrVZrM/ieOHGCNWvW8OOPP+Lg4MC2bduoW7cuBw4cICQk5JXWthWH4OBgLl26xOzZs/Otc+rUKYyMjLSjXQVR1q+lKJic1zEyMpJBgwblmrqaU2RkJB06dGDnzp20bduWcePG0bBhQ1auXMmECRPQ09MjIyODb775hgMHDmBpacnAgQMZPHgwHh4evP3221y8eBEjo/Kx56z6xwA0Z4/lKddzcEF/iL+OI0qXfCYrBrmOFYdcy5dTu3ZtgDx3NWUKrhDipcUmpRHwazRTlLcI+DWa2KS0Fx9Ugp4dybG0tHxu/YJuD1GWnT59mvPnz5d2N0Q5kD1C6uLiQkJCAn369CE5OZlr164RFRXF3bt36dixI/369cPNzQ07Ozs8PDwA8Pb2RqFQMGrUqFI+i4LRPIorVLkQQojiI1NwhRAvJTohhWmHIolNyl7Mn8KVByllah/K7DVwbdu2ZceOHdjZ2bF+/XoiIiJ466232LZtGx06dMhzXJs2bZg0aRK3b9/WTsGtVq3aS/cjMjISb29v7OzsCAsLo3HjxixZsoRz584xc+ZMMjMzsbe3Z+7cuRgbG+Pg4MDevXuxsrLi4sWLzJw5k61bt+Zq8+HDh3z55ZdERUUBMH36dKytrVm3bh0GBgZs27aNWbNmFWpbC1FxPJu4R9frx48fz1W2f/9+hg8fnqvsrbfeomvXrtrAM9vkyZOZPHly0XW4mOlVtULXfC+9qlY6SoUQQhQnGQEVQryUFadv5wg+s8QmpbPhYtmZovLsCM+wYcNYuHAhw4cPx9XVFX19fXx8fPIcV716debPn8+QIUNQKBSMHDnylfvy7J6My5cvx8/PT7s/a0ZGRqG2g5g6dSpDhw5lz549/PDDD4wbNw4bGxt8fHwYOnQoSqVSgk9RKElJSTrLVSoVqqRMLpxWcepwIhdOZz0vV3p7w7NrPWtaZ5ULIYQoUTICKoR4KQ9VT3SWx6ek6ywvDYaGhnz77bdAVtbgL774gh9++IEDBw7kqZu9TUS2Ll260KVLF53tnjp1iqCgoEIFjM/uybho0SLq1auHra0tAJ6enqxZs0abEOlFTpw4wZUrV7TPk5KSUKl0bHAmRAGZm5vrLDeuZMqZo0kkq7LHEDOJf5hBh07mmJmXj+2K9Gtao/abATs3oHkUlzXy2dtbEhAJIUQpkABUCPFSqpvpTltezbRsJiWxtrbmhx9+eG4dVVIm4WGppKaoMTHVp0lLkyL7gq1rPWp+2XUNDQ1Rq9UA2m1inqVWqwkNDcXExKRI+idEhw4diImJ4fHjx9oyCwsLqpq1Iu5e7gmsySoN4WGptH7XrKS7+dL0a1pDKSccWrlyJWvXrqVly5YFSnIWGxvLV1999cLfXUIIUZ7IFFwhxEsZ9m49rM1zB5vW5kZ429copR79Y86cORw6dEi7Bi4gIICgoCDtiGZkZCR9+/bFw8MDDw8Pfv/9d1RJmaxefpixE7yZ881ohvzHg0E+I0lKzNpL9ciRIzg7O+Ph4cHevXu175WcnMzYsWPp3r077u7u7N+/X2efnt2T0c7OjsjISCIiIgByrUetW7culy5dAmD37t0623NxcWH16tXa55cvXwbAzMws36mUQjyPpaUlffv2pUmTJtStW5cmTZrQt29f9DS6g8zUVHUJ97D8W7NmDRs3bswVfD5vv+aC3DgTQojyRgJQIcRLqW1pynRXG5z/ZUHLN0xx/pdFmUlA1KtXL0JDQ7XPQ0NDadWqlfZ5jRo12LhxI/v372fZsmVMnTo1a+QzVUNk1N949fmSaeNDiL17h1+2/Epqair//e9/+emnn9i3bx/37t3TtrV48WIcHR3ZvXs3W7ZsYebMmSQnJ+fpU2HWo44dO5apU6fSrVs3DAx0j8DOnDmTixcvolAo6NSpE+vWrQPAzc2Nffv24ebmlmdasRAvYmlpiYeHB/369cPDwwNLS0tMTHV/VTAxka8QhTFhwgRu376Nj48PTZs2ZfTo0fTu3Zsvvvgiz02x06dPA1k3y7JvnIWHh9O9e3fc3NxQKBTcuHGjNE9HCCFemkzBFUK8NGvzSvg71i7tbuTRokULHjx4QGxsLA8fPsTS0jJ7LyoA0tPTmTx5Mn/99Rf6+vrcuHGD1JSs0Zx/1WtJtapZ68Js6jQlKuoO165Vp169ejRo0ACADz74gPXr1wNw/PhxlEolQUFBQNaU2aioKBo1apSrTznXo2br2LGjzvWoDg4OnDx5Mk/5gAEDGDBgAABWVlba98zJ1taWgwcPFuwHJUQBNGlpQvzDjBxrQKGymR5NWsr078KYN28eR48eZcuWLaxevRqlUskvv/yCqakpKSkpbNy4ERMTE27cuMGYMWNy3UQDWLduHZ999hn9+vUjLS2NzMxylghKCCGekgBUCFEh9ejRg927d3Pv3j169eqV67UffviBmjVrolQqUavVNGjQQDvKY2T4zwiunp4++gbPn2ao0WhYsWIFDRs2LPqTKICkJ/cIu7+VlPRHmBpVpWXN/pgb1yqVvoiKyczcgA6dzJ/OElBjYlK066NfV+7u7piamgJ5b4plT83PqU2bNixZsoSYmBi6deumvSEmhBDljcyfEUJUSL169WLnzp3s3r2bHj165Hrt8ePH1KpVC319fbZt20ZmZiZNWppgYpI7UZCRkR5v1jWiYcOGREZGcvPmTSBrDWe27LWYGk3W6FD2WsycXrQn48tKenKPo7fmcTvhNPeT/+Z2wmmO3ppH0pN7Lz5YiEIwMzeg9btmvNe5Cq3fNZPgswhUrlxZ+zjnTbG9e/eSlpaWp37fvn1ZvXo1JiYm+Pj46JwlIYQQ5YEEoEKICqlJkyaoVCqsra154403cr02aNAgtm7dikKh4Nq1a1SuXBkzcwPefscUYxM9qtcyoE49I96oY0QlYz1MTEyYP38+H3/8MR4eHtSo8U+iJV9fX9LT01EoFHTu3Jn58+eX2DmG3d+KKj13sKlKzxoRFUKUH7puij3r1q1b1K9fn88++wwPDw/+/vvvYumLj48PCQkJxdK2EEKATMEVotRlp+VPSkqia9euzJ49O9+6Bw4c4MqVK4waNapQ7+Hg4MDevXuxsrJ61e6WK4cOHdI+zjkK2aBBg1zrJCdPngxAF1cnurg6actbvztH+7hz58507tw5z3uYmpqWaNCZU0r6I53lqfmUCyHKpkGDBjFs2DC2bt1K586dMTPLm3k4NDSUbdu2YWhoSK1atRg9enSx9CU7oVlBZWZm5pssTRSta9euMXLkSPT09FixYgV//PEHffv2Le1uCVFoetnTxkqYJjo6ulAH1KhRgwcPHhRTd0RJkeuYl7OzM5s2beLEiRNcunTpuQHoyyqOAFSuZfHuG1oQp+98z+2E03nK61m+y7t1/1PgduRaVgxyHSuOkrqW27ZtY9WqVaSlpdGqVSvmzp3Le++9p/17oet1AwMDGjVqxEcffcSJEyeYM2cO7du3L/a+lkdFfR2XLl1KRkYGvr6+nDp1iqCgINauXVtk7Yv8ye/Xl/M0AaTes+UyAipEKcqZlj87uynAw4cP+fLLL4mKigJg+vTptGvXjuDgYG2QGhoaSmBgIPr6+lhYWLB9+3YyMzOZPXs2R48eRV9fn4EDBzJ48GAAVq1ahVKpJCMjg+XLl5da0pyKQpWUyZmjSTkyg2YS/zCDDp3MSywIbVmzPw+Tr+eahmtmVIuWNfuXyPsLIYqf+n4s7NyA5lEcelWtoLc3+jWtX7ndq1evEhISwo4dOzAyMmLixIls3779ha97enqSnJxMq1atmDZt2iv343WXnJzM8OHDiYmJQa1WM2bMGG7cuIFSqSQ1NZW2bdsyb948Dh8+zI8//oiBgQEnT54kNTWVa9eu4ebmhqenJ8OGDSvtUxGiwCQAFaIU5UzLr1QqteVTp05l6NChtG/fnqioKAYOHMixY8dyHbto0SI2bNjAm2++qV2vs379eiIjIzlw4ACGhobEx8dr61tZWbF//35++ukngoKCWLBgQcmcZAUVHpaaa1sKgGSVhvCwVFq/m3f6XHEwN65Fp/oTCLu/ldT0R5hIFlwhKhT1/Vg0gVPhfiwAGoAb4aj9ZrxyEHry5EnCwsJ4//33AUhNTc21vv15rxsYGNC9e/dXen+R5ciRI1hbW2unPj9+/BhnZ2f8/PwAGD16NEqlEnd3d3x8fDAzM2PEiBEyAirKNQlAhSiDTpw4wZUrV7TPk5KSUKlUueq0bdsWPz8/evbsSbdu3YCsLww+Pj4YGmZ9tKtVq6atn13Hzs6OvXv3FvcpVHjZ+4bmKU99/rYtRc3cuFahptsKIcqRnRu0wafW0xFRhvi/UtMajQZPT08mTpyYq3zz5s3PfR3A2NhY1n0WkaZNmzJjxgxmz56NQqHAwcGB3bt3s2zZMlJSUnj06BFNmjTB3d29tLsqRJGRLLhClEFqtZrQ0FCUSiVKpZLz58/nSUoxb948xo8fT3R0NN26dSMuLu65bRobGwNZd65lA/NXl71vaJ5yE/m1Wh7179+fixcvvnIdIYqS5pHu3+v5lReGk5MTu3bt0q5ri4+P586dOwV+XRQNW1tb9u3bR9OmTZk/fz6BgYFMmjSJ5cuXc+jQIQYOHMiTJ09Ku5tCFCn5piREGZS9t2Q2XXtL3rx5k9atW/Pf//6X6tWrEx0dTceOHVm3bh0ZGRkAuabgiqLVpKUJlc1yr6uvbKZHk5YmpdQjIURFo1dVd+K4/MoLo3HjxowfP54PP/wQhULBhx9+yN27dwv8uigasbGxmJqa8sEHHzBixAjCwsKArGUzKpWK3bt36zzO3Nw8z8woIcoLmYIrRBk0c+ZMJk2ahEKhICMjAwcHB+bNm5erzqxZs4iIiECj0eDk5ETz5s1p2rQpN27cQKFQYGhoiLe3N59++mkpnUXFZmZuQIdO5llZcFPVmJiUfBZc8Y/IyEg++ugj2rdvz7lz57C2tmbVqlX4+Pjw1VdfYW9vT1xcHN26dePs2bOkpKQwduxY/vrrLxo2bEhqaqq2rWPHjrFgwQLS0tKoX78+gYGBeWYgNGrUiM8++4yDBw9iYmLC6tWrqVmzZkmftqjoenvDjfDc03BrWmeVF0XzvXvTu3fvXGVnz5597uuQlaBIFI3/+7//Y9asWejp6WFkZMTcuXPZt28frq6u1KxZE3t7e53HNWvWDH19fRQKBV5eXpKESJQrr7wNi5eXlw2wFniDrPXxKzZv3rz4BYfJNiyvKbmOFYdcy4qjIlzLyMhIHB0d2bNnDy1atGD48OG4u7uzceNGnQHo8uXLCQ8PZ+HChfz111907dqV0NBQbGxsGDJkCOvXr6dy5cp89913pKWl4efnR//+/bVt1alTh9WrV+Pu7s6sWbMwNzfH19e3VH8GFeE6iiw5r2VxZcEtrNLedqo8KurPZFn5v/A6kt+vL6c4t2HJAPw3b958wcvLqwpw3svLS7l58+a/iqBtIUQRiE1KY8PFB8Qnp1OtshHe9jWwNq9U2t0SokjZ2NjQokULICvZVmRkZL51z549q92i6O2336ZZs2YAnD9/nitXrmhHfdLT02nTpk2e4ytVqoSbmxsALVu25MSJE0V6LkJk069p/coJh15VWdh26nVXnBmRhShprxyAbt68OQaIefo40cvL62+gDiABqBBlQGxSGtMORRKblP60JIUrD1KY7mojQWg5M27cOIYNG0bjxo1Luyul7uLFi2zdupWZM2cSHBzMqVOnMDY2JiAgADMzMwwMDEhNTcXAwAC1Oiszcc5ptvnRaDQ4Ozvz/fffP7eeoaEhenpZN3UNDAy0666FqIjKwrZTr71izIgsREkr0jWgXl5e/wJaAWd1vDYMGAZZKb5z7jVVEIaGhoU+RpQ9ch1L3tJz4TmCzyyxSels/b9E/l/XJi/drlzLkvfTTz8VS7vl8Vq6urri6uoKQJUqVbTbQlSuXJnKlSsDWcFko0aNuHHjBm5ubvz8888YGBhQo0YNunTpwr59++jTpw9//vknf//9N1WrVsXOzo6pU6fy6NEjGjZsiEqlIioqisaNG2NkZETVqlWpUaMGenp62p+ZhYUFJiYmpf4zLI/XUehW1q5lZmYqkK6j3KBM9bOsKcrrGKdK1HEFwFCViJVcg2JX1j6T5V2RBaBeXl7mwDbAd/PmzY+ffX3z5s0rgBVPn2oKO49a5l5XDHIdS15sfJLu8kdJr3Qt5Fq+usjISLy9vbGzsyMsLIzGjRuzZMkSzp07x8yZM8nMzMTe3p65c+dibGysXYPYokUL/P39uXTpEnp6egwYMIBhw4axcuVK1q1bh6GhIY0aNWLZsmXEx8fj7+/P7du3MTExYf78+bz99tsEBAQQFRXF7du3iYmJYfDgwXz22Wclcs66kgVdv36dL7/8ktTUVOrXr09AQABVq1bNte4y5xrOnJuwJyYm8uTJEzIzM0lOTtaOTEZHR/Pnn3+yceNGvvzyS/r160dmZiarV6/mxx9/5O7du2zcuJHOnTtjZ2fHo0ePqF+/PgsWLGDgwIGkpaUBMH78eKysrEhPT+fRo0c8ePAAjUaj/f//+PFjUlNTS/3zIJ/JiqOsXUsDA91bdxkYZJapfpY1RXkd1WZVdJZnmFWRa1ACytpnsrx4ugY0jyIJQL28vIzICj43bN68eXtRtCmEKBrVKhsBKXnLTY1KvjMij+vXrxMQEEC7du0YO3Ysy5cvZ/369QQHB2Nra8sXX3zB2rVrGTp0qPaYP//8k9jYWA4fPgxAQkICAN999x2nT5/G2NhYWxYQEECLFi1YtWoVJ0+eZMyYMSiVSgCuXbvGli1bMDY2pnnz5nz88ccYGRX//4uIiAi+++47vvnmG4YPH86ePXtYtmwZM2fO5N133+Wbb75h4cKFzJgxo8Btmpubc/jwYQICAgAYMWIEXl5eBAQE0KBBAy5cuMDcuXM5e/Ysrq6ubNy4kTfffJOEhAQsLS1zteXk5MSePXvyvMfWrVu1j3NmAe3Rowc9evQo7I9BiHKjSUsT4h9m5JqGK9tOlbBizogsREl65X1Avby89ICVwN+bN29e+OpdEkIUpayEQ7mDCmvzrEREovTVrl2bdu3aAdCvXz9OnjxJvXr1sLW1BcDT0zPXtggA9erV4/bt20yZMoUjR45QpUrWnfFmzZoxatQotm3bhqFh1v3F3377jQ8++ADICqzi4+NJTEwEsqaxGhsbU6NGDWrUqMH9+/dL5JyfTRZ069YtEhISePfdd/M958JSqVScP3+e4cOH4+bmxoQJE7h37x4Abdu2xc/Pjw0bNpCZqXtkJz/q+7Gofwwgc8Fk1D8GZGWlFKKCy952qk49I6rXMqBOPSNJQFTC9Gtao+c3Az0HF2jSEj0HF/QkAZEop4piBNQR8AHCvLy8/ve0bNLmzZvz3j4WQpQ4a/NKTHe1ycqCm5JONVPJgluWZE8XzWZpaUl8fPxzj6latSpKpZKjR4+ybt06QkNDWbhwIWvXruXMmTMolUqWLFnCoUOHntuOsbGx9rGBgUGhg7GX9ez7Zo/W6lLYJELZ1Go1FhYW2tHenObNm8eFCxc4dOgQ3bp1Y+/evVhZWb24TclCKV5jZuYGknColJWFjMhCFIWiyIJ7Eh37uwghyg5r80r4O+qehy9KV1RUFOfOnaNt27bs2LEDOzs71q9fT0REBG+99Rbbtm2jQ4cOuY6Ji4vDyMiI7t27Y2try+jRo1Gr1URHR+Po6Ej79u0JCQlBpVLh4ODA9u3b8fPz49SpU1hZWWlHTMsKCwsLLC0tOXv2LA4ODrnO2cbGhkuXLtGqVSt2795d4DarVKmCjY0NoaGh9OzZE41Gw19//UXz5s25efMmrVu3pnXr1hw5coTo6OgCBaCShVIIIYR4dUWaBVcIIUTh2NrasmbNGvz9/WncuDEzZ86kdevWDB8+XJuEyMfHJ9cxMTExjB07VjsyOHHiRDIzMxk9ejSJiYloNBoGDx6MpaUlY8eOxd/fH4VCgYmJCYsWLSqN08xXaGgo9vb2LFq0iJ49e1K/fn0aNGjA8ePHmT59OiNGjGDEiBFs2LBBm/W2oJYuXcrEiRNZvHgxGRkZ9O7dm+bNmzNr1iwiIiLQaDQ4OTnRvHnzArWneRRXqHIhhBBC5KWn0WheXKvoaaKjowt1gGSfqhjkOlYcci1fXWRkJIMGDdImEyotZeVaOjg4aKfDNmrUKFein7JA/WMAmrPH8pTrObigXwZGQMvKdRSvTq5lxSDXseKQa/lynmbBzTNT9pWTEAkhhChfDJ48oeqt21S/eoOqt25Dct4syTlt2bIFhUKBQqFg9OjRREZG4unpiUKhFPJKPwAAIABJREFUwMvLi6ioKABu375Nz549cXV1Zd68eTRq1AiAU6dO8fHHH2vbmzx5MsHBwQD079+fixcvPvf9ly1bxvvvv49CoWDBggUAJCcn4+Pjg0KhoEuXLuzcufOF5x2blEbAr9FMUd4i4NdoYpPSXnhMLr29s7JO5iRZKIUQQohCkSm4QghRSmxsbEp89NPgyROsrkdglPZ0S3MVaP64iMG/bMjMkRwoW3h4OIsXLyYkJAQrKyvi4+Px9fXF09MTLy8vNm3axFdffcWqVauYOnUqH3/8MZ6envz0009F0t9jx44RERHB7t270Wg0fPLJJ5w5c4aHDx9ibW3NunXrgKy9OJ8nNimNaYciiU3K3so9hSsPUpjualPghFz6Na1R+82AnRvQPIpDr6oV9PaWBERCCCFEIcgIqBBCvEaqxN79J/h8Si8lhSqxd3XW//XXX+nRo4c2SU+1atU4f/48ffv2BeCDDz7gt99+A+D333+nT58+2vKicOzYMY4dO4a7uzseHh5cv36diIgImjZtyvHjx5k9ezZnz57FwsLiue1suPggR/CZJTYpnQ0XCzelSr+mNfpD/DEYNxv9If4SfAohhBCFJCOgQgjxGjFIyyhUeWE9u60MgKGhITnzDTx58qTA7Wk0GkaNGpUnERPAvn37OHz4MPPnz8fJyQk/P79824lPTtddnqK7XAghhBDFQ0ZAhRDiNZJZSfd9x/zKHR0d2bVrF3FxWZle4+Pjadu2rXbN5fbt23FwcACgXbt2ucqz1alThytXrvDkyRMSEhI4efJkgfvbqVMngoODUalUQFYG4AcPHhAbG4upqSkffPABI0aMICws7LntVKtspLvcVHe5EEIIIYqHjIAKIcRrJNH6DYxUybmm4WpMTUm0fkNn/SZNmvDFF1/Qv39/9PX1adGiBbNmzcLPz4+goCCsrKwIDAwEYMaMGXz++ed8//33uLu7a9uoU6cOPXv2pEuXLtSrV48WLVoUuL8uLi5cvXqVXr16AVC5cmW+/fZbbt68yaxZs9DT08PIyIi5c+c+tx1v+xpceZCSaxqutbkR3vY1CtwXIYQQQrw62YZFlCi5jhWHXMvyy+DJE6rE3sUgLYPMSoYYNmvGg2RVkb9PWdtKJTYpjQ0XHxCfkk4106zgs6AJiMoD+UyWTw8fPmTQoEGkpaUxc+ZMHBwc5FpWEHIdKw65li8nv21YZARUCCFeM5nGxjyqX0/7vEZlUyiGALS4JD25R9j9raSkP8LUqCota/bH3LjWC4+zNq+Ev2PtEuihEAV38uRJmjZtqt1iSAghKjpZAyqEEKJYFMfoZ9KTexy9NY/bCae5n/w3txNOc/TWPJKe3Cvy9xLiRZYvX06XLl3o0qULP/zwAwDbtm2je/fuuLm5MX78eDIzMwG0++IC7Nq1C19fXy5fvsysWbPYv38/bm5uJCcn4+vrS6tWrXB1dWXFihWlcl5CCFGcJAAVQpRZOb+wFcS+ffu4cuVKMfVGlAVh97eiSs8dbKrSs0ZEhShJly5dYvPmzezatYvQ0FB+/vlnzp07R0hICDt27ECpVGJgYJArIdezWrRowbhx4+jVqxdKpZJr164RGxvLH3/8waFDhxgwYEAJnpEQQpQMmYIrhKgw9u3bh0KhoHHjxqXdFVFMUtIf6SxPzadciOLy22+/0bVrVypXrgxAt27duHjxImFhYbz//vsApKamUqNGwRNd1atXj9u3b+Pr64ujoyMuLi7F0nchhChNEoAKIcqEwYMHEx0dzZMnT/jss8/46KOPAJg2bRrHjx+nZs2aLFu2jOrVq7NhwwY2bNhAWloab731FkuWLOHy5csolUrOnDnD4sWL+eGHHzh06BDr1q3D0NCQRo0asWzZslI+S/GqTI2q6iw3yadciJKk0Wjw9PRk4sSJeV7LuUdufnvhVq1aFaVSyYULF1i9ejWhoaEsXLiw2PorhBClQabgCiHKhICAAPbt28eePXtYtWoVcXFxJCcnY29vz5EjR3j33Xe1X8S6devGnj17OHjwIA0bNmTjxo20a9cONzc3pkyZglKp5F//+hffffcd+/fv5+DBg3z99delfIaiKLSs2R8zo9wJh8yMatGyZv9S6pF4XTk4OLB//35SUlJITk5m37592Nvbs2vXLm22zPj4eO7cuQNAzZo1uXr1Kmq1mn379ulsMy4uDrVaTd++fRk/fvwL97etCFauXImLiwujRo0qlvaDg4OZPHmyzteWLFlSLO8phHg+GQEVQpQJq1atYu/evQBER0cTERGBvr6+dv/Hfv36MWTIEADCw8OZP38+jx8/RqVS5TtNrVmzZowaNYquXbvStWvXkjkRUazMjWvRqf4Ewu5vJTX9ESaFyIIrilZAQABmZmaMGDGiSNq7fPkyd+/exdXVtUjaK24tW7bE09OT7t27A/Dhhx/Srl07xo8fz4cffohGo8HQ0JDZs2dTt25dJk6cyKBBg7CyssLe3h6VKm/m6ZiYGMaOHYu+vj4ZGRk6R1IrmjVr1rBp06bs7RoAyMjIwNCw+L+ifvvtt3zxxRfF/j5CiNwkABUih5UrV7J27Vru37/Pf/7zn0LfkXVwcGDv3r1YWVkVUw8rplOnTnHixAlCQ0MxNTWlf//+OqeoZU9h8/PzY+XKlTRv3pzg4GBOnz6ts921a9dy5swZlEolS5Ys4dChQyXypeZ1sm/fPho0aKBddxscHIyLiwvW1tbF9p7mxrV4t+5/iq19UTr+/PNPLl26VKIBaEH2qu3VqxchISFERkZy7tw5+vbtC8DFixe5c+cOhw8fzlW/d+/e9O7dO087PXr0oEePHnnKBwwYoE021Lx5c/bv3//a7Dk4YcIEbt++jY+PD1FRUbi5uXH79m3q1KmDi4sLly5dYvbs2QB8/PHHjBgxgvfee49jx46xYMEC0tLSqF+/PoGBgZiZmfG///2PqVOnkpycjLGxMcHBwQDcvXsXb29vbt68Sbdu3ZgyZQpz5swhNTUVNzc3mjRpwtKlS9m2bRurVq0iLS2NVq1aMXfuXAwMDErzRyREhSRTcIXIYc2aNWzcuJG//vqr2KYDibwSExOxtLTE1NSUa9euceHCBQDUajW7d+8G4JdffqF9+/YAJCUl8cYbb5Cens4vv/yibcfc3Fw7qqBWq4mOjsbR0ZHJkyeTmJioc8ThdZSRkVFkbT2beXjLli3cvXu3yNoXZcvixYtxcnKiT58+XL9+HYCbN2/i7e1N165d6du3L9euXQPA19eXCRMm0K1bN5ycnFAqlUBWYh4/Pz9cXV1xd3fn119/JS0tjQULFhASEoKbmxs7d+4stXN8VkhICACRkZG5ft/Y29szc+bMArfzos9dQkIC+/fvZ9u2bWzdupWEhISX63A5Mm/ePN544w22bNnC0KFDuXr1Kps2beL777/P95i4uDgWL15McHAw+/fvx97enhUrVpCWlsbIkSOZMWMGBw8eZNOmTZiYmABZNzeWLVvGoUOHCAkJISoqikmTJmFiYoJSqWTp0qVcvXq1UBmMhRAvT4YChHgq553YAQMGcOvWLWbPnk1oaCiBgYHo6+tjYWHB9u3byczMZPbs2Rw9ehR9fX0GDhzI4MGDgayppEqlkoyMDJYvX07Dhg1L+czKvk6dOrFu3TpcXFywtbWldevWAFSuXJk//viDxYsXU716dYKCggD473//S48ePahevTqtWrUiKSkJyBp5+O9//8vKlSv5/vvvGTduHImJiWg0GgYPHoylpWWpnWNJ27JlC8uXLweypiIbGBhgbGzMn3/+Sdu2bfnkk0+YPHkyDx8+xMLCgjlz5lCrVi0UCgVnzpxBX1+f5ORknJ2dOX36NFFRUdr6pqamfPPNN8THx+dK/NSnTx8uXrzIqFGjMDExISQkhKtXrzJ9+nRUKhVWVlYEBgbyxhtvsHLlSkkQVc5cunSJkJAQ7e+3rl27Ymdnx/jx4/n6669p0KABFy5cYOLEiWzZsgWAO3fusHv3bm7evImnpycdO3bkp59+Qk9Pj0OHDnHt2jU+/PBDTpw4wbhx43KNeJUklUrFp59+SkJCAhkZGYwfPx4PDw/gn1HSOXPmcO3aNdzc3PD09KRFixYEBQWxdu1akpOTmTJlCuHh4aSnp+Pv74+HhwfBwcHs3bsXlUqFWq3m+++/Z+TIkSQmJpKZmcncuXNxcHAgISGBX375hcePHwMQFRXFzZs36du372v1e8vd3R1TU9Pn1jl//jxXrlzRjjKnp6fTpk0brl+/Tq1atXjnnXcAqFKlivYYJycnLCwsAGjcuDFRUVHUqVMnV7snT558pQzGQoiCkwBUiKfmzZvH0aNH2bJli/ZOPcCiRYvYsGEDb775pvaO9Pr164mMjOTAgQMYGhoSHx+vrW9lZcX+/fv56aefCAoKYsGCBSV+LuWNsbEx69evz1Oe39S4QYMGMWjQoDzl7dq14+jRo9rnO3bsKLI+lifh4eEsXryYkJAQrKysiI+PZ/r06cTExLBz504MDAzw8vLSBg03btxgwoQJbNmyhebNm3P69GkcHR1RKpV06tQJIyOjfIMMNzc3FAqFdmrhkSNH+Oqrr7C3tyc9PZ0pU6awevVqqlevzs6dO5k3bx4LFy7ku+++4/Tp0xgbG78WIz0VwdmzZ+natas2QHBzcyM1NZXz588zfPhwbb20tDTt4549e6Kvr0+DBg2oX78+165d4/fff+fTTz8FoGHDhtStW5cbN26U7Mk8w9jYmJUrV1KlShXi4uLo2bMn7u7uuTLXTpo0SRtwQtbSgWyLFy/G0dGRhQsXkpCQQPfu3enYsSMAYWFhHDx4kGrVqhEUFISLiwtjxowhMzOTlJQUAM6cOaMNPrM9fvyYM2fOaAPh10H2ljYAhoaGqNVq7fPsZRkajQZnZ+c8o6R///13vu1WqlRJ+zh7fe2znpfBWAhRtCQAFeIF2rZti5+fHz179qRbt25A1p1SHx8f7XrCatWqaetn17Gzs9Mm1RElKyEhgTNnzpCUlIS5uTkdOnR4rUYRfv31V3r06KFdi5z9/7NHjx4YGBigUqlyBQ2GhoYkJycD/6x3c3R0JCQkhEGDBuWpD7mDjPxcv36d8PBw/v3vfwNZ06Jr1cpKFvS6JYjy8fFh6dKlFe7/oUajwcLCItdNu5xyBnC6npcVGo2Gr7/+mrNnz6Knp0dsbCz379/X/n99kePHj6NUKrWzNJ48eUJUVBQAzs7O2s/gO++8g7+/PxkZGXh4eNCiRQsA7SyOZ73OywZsbGxYs2YNarWamJgY/ve//wHQpk0bJk+eTEREBG+99RbJycnExMRga2vLvXv3+N///sc777xDUlKSdgpufoyMjEhPT8fIyAgnJyc+/fRThg4dSo0aNYiPj0elUlG3bt2SOF0hXisSgArxAvPmzePChQscOnSIbt26vTCoNDY2BsDAwIDMzMyS6KLI4dmpbJCVWfJ1m8qmS/boglqtzhU05Ex44u7uztdff018fDyXLl3C0dGR5OTk5wYZ+dFoNDRu3JjQ0NA8r71OCaI0Gg1r1qxBX798p13o0KEDfn5+jBo1iszMTJRKJT4+PtjY2BAaGkrPnj3RaDT89ddfNG/eHIBdu3bh6enJ7du3uXXrFra2trRv355ffvkFJycnrl+/TlRUFLa2tkREROQbiBW37du38/DhQ/bu3YuRkREODg757tWpi0ajYcWKFXmWXFy4cCHXqF6HDh3Ytm0bhw4dws/Pj2HDhuHp6Ym5ubnOds3MzF7uhCqAdu3aUa9ePTp16kSjRo1o2bIlANWrVycwMJDPP/9ceyNs/Pjx2NrasmzZMqZMmUJqaiomJibaJET58fb2RqFQ0LJlS5YuXZpvBmMhRNGqmH/thShCN2/epHXr1rRu3ZojR44QHR1Nx44dWbduHe+99552Cm7OUVBRemQqGzg6OvLZZ58xbNgw7RTcnKpUqZInaPjzzz9p3rw5ZmZm2NvbM3XqVBQKBQYGBjrrZwcZORM/QdYX5uwgwtbWlri4OM6dO0fbtm1JT0/nxo0bNGrUSJsgqn379oSEhKBSqSrUDYLIyEgGDhxIq1atCAsL48qVK4SFhbFr1y7WrVsHZCXfqlu3Llu3bi3l3hZMy5Yt6dmzJ25ubtSoUUO71m7p0qVMnDiRxYsXk5GRQe/evbUBaO3atenevTuJiYl8/fXXmJiYMGjQICZOnIirqysGBgYEBgZibGzMe++9x3fffYebmxujRo3SmUm2uCQmJlKjRg2MjIz49ddftXt35vTs//WcXFxcWL16NbNmzUJPT4/Lly9rRzdzunPnDm+++Sbe3t6kpaURFhaGp6cnHTp0ICYmJtfvLgsLCzp06FB0J1lGnT17FgB/f/9c5Xp6eixdulTnMU5OTuzZsydP+TvvvMOuXbtyleXMMgxop1ADTJ48OdceofllMBZCFC0JQIV4gVmzZhEREYFGo8HJyYnmzZvTtGlTbty4gUKhwNDQEG9vb+2aJlG6ZCobNGnShC+++IL+/fujr6+v84twzqBBo9HQo0cPbdDQq1cvhg8fniswyi/IyJn4acWKFXh5efHll19qkxAtX76cqVOn8vjxYzIzMxkyZAgNGjRg9OjRFT5BVEREBIsWLaJNmzY4ODgAWVtJfPzxx6Snp+Pl5cWwYcNKuZeFM2bMGMaMGZOnfMOGDTrrd+zYkXnz5uUqMzExITAwME/datWq6QwqSkK/fv0YNGgQrq6u2NnZ6Uwe16xZM/T19VEoFHh5eeX6XPn6+jJt2jQUCgVqtRobG5tcgU62U6dOERQUhKGhIWZmZixevBgAS0tL+vbty5kzZ7QJu1q1alUhPxdlVdKTe4Td30pK+iNMZX9hIYqVnkajKY331URHRxfqgNdlT6yKTq5jxVFWr+X+/fsJDw/PU96kSZPXZgS0sMrqtSzPIiMj8fT05MyZM0DePYInTpxI9erVGTduXJG9Z1m7jr6+vrkSVOkSm5TGhosPiE9Op1plI7zta2BtXinf+q+LsnYtK7qkJ/c4emseqvR72jIzo1p0qj/hlYJQuY4Vh1zLl1O7dm2APIv/ZQRUiCIid0/Lhtd5Klt5oJ/2ELM4JQYZiWQaVkFl5Ya6UvXS7laxyLn2L6fg4GDu3LlTKtuNlKRFixY99/XYpDSmHYokNin9aUkKVx6kMN3VpsIFoer7sbBzA5pHcehVtYLe3ujXtC7tbomnwu5vzRV8AqjSs/6mv1v3P6XUKyEqLglAhSgCuu6ePky+/sp3T0XhPTuVzczM7LXLgltW6ac9pGr0Kgwz4rRlRqmRPKo9uMIGoc+6dOkSy5cvZ/v27eU+KdGr2nDxQY7gM0tsUjobLj7A37F2KfWq6Knvx6IJnAr3YwHQANwIR+03Q4LQMiIl/ZHO8tR8yoUQr0YCUCGKgNw9LVssLS1lum0ZZBanzBV8AhhmxGEWpyTR+t+l1KuStXr1ah49eoSnpycA9vb2r+1ewfHJ6brLU3SXl1s7N2iDT62nI6IM8dd9jChRpkZVdZab5FMuhHg1EoAKUQTk7qkQL2aQkVio8vLMxsaGw4cPa59nZ/rUlXzndVWtshGQkrfc1KjkO1OMNI/iClUuSl7Lmv15mHw9zxrQljX7l2KvhKi4JAAVogjI3VMhXizTsEqhyisSWSOel7d9Da48SMk1DdfaPCsRUUWiV9UKXeke9apalXhfhG7mxlkJh8LubyU1/REm8hkVolhJACpEEZC7p0K8mMrKDaPUyFzTcDMMrVBZuZVir4qfrBHXzdq8EtNdbbKy4KakU820gmbB7e0NN8JzT8OtaZ1VLsoMc+NasmRGiBIiAagQRUDungrxYupK1XlUe/BrkwU3m6wRz5+1eaUKlXBIF/2a1qj9ZkgWXCGEeEoCUCGKiNw9FeLF1JWqvzYJh7LJGnGhX9NaEg4JIcRTr3cOeCGEEKKYyRpxIYQQ4h8SgAohhBDFqGXN/pgZ5Z6OL2vEhRBCvK5kCq4QQghRjGSNuBBCCPEPCUCFEEKIYlaQNeK9evUiJCSkhHokhBBClA6ZgiuEEEKUEI1Gg1qt1vmaBJ9CCCFeBxKAClEG9erVq9Te++LFi3z11Vel9v5CVDSRkZF07NiRL774gi5durBo0SLef/99FAoFCxYs0NZr1KgRAHfv3qVfv364ubnRpUsXzp49W1pdF0IIIYqcTMEVogwq7pGQjIwMDA11f/zt7e2xt7cv1vcX4nUTERHBokWLSEpKYvfu3ezevRuNRsMnn3zCmTNn6NChg7buL7/8gouLC2PGjCEzM5OUlJRS7LkQQghRtGQEVIgyKHsk5NSpU/Tv35+hQ4fi7OzMqFGj0Gg0ADg4OLBgwQI8PDxwdXXl2rVrACQnJzN27Fi6d++Ou7s7+/fvByA4OJhPPvkET09PBgwYkG+9U6dO8fHHH7+wLS8vL7y9vXF0dGTWrFkl+vMpapmZmcXSbkZGRrG0K8qfunXr0qZNG44dO8axY8dwd3fHw8OD69evExERkavuO++8w+bNmwkICODvv//G3Ny8lHothBBCFD0ZARWijLt8+TKHDx/G2tqa3r178/vvv9O+fXsArKys2L9/Pz/99BNBQUEsWLCAxYsX4+joyMKFC0lISKB79+507NgRgLCwMA4ePEi1atWYO3duvvWyPa+tixcvsnfvXipVqoSzszOffvopderUKdkfTgFERkbi7e2NnZ0dYWFhNG7cmCVLltCpUyd69erF8ePH+c9//kPVqlVZsGABaWlp1K9fn8DAQMzMzHBwcKBnz54cOXIEExMTli5dyltvvcWBAwdYsmQJaWlpVKtWjaVLl1KzZk0CAgK4efMmt2/fpk6dOsycOZMvv/ySqKgoAKZPn067du04ffo0U6dOBUBPT4/t27dLoJHD80bpy6PKlSsDWWtAR40ahY+PT751O3TowLZt2zh06BB+fn4MGzYMT0/PkuqqEEIIUawqzl93ISqod955h9q1awPQvHlzIiMjtQFot27dALCzs2Pv3r0AHD9+HKVSSVBQEABPnjzRBj/Ozs5Uq1bthfWyPa9Oly5dsLCwAKBx48ZERUWVyQAU4Pr16wQEBNCuXTvGjh3LmjVrAKhWrRr79+8nLi6OIUOGEBwcTOXKlfnuu+9YsWIFfn5+AFSpUoVDhw6xZcsWpk2bxtq1a2nfvj2hoaHo6enx888/8/333zNt2jQArl69yi+//IKpqSmff/45Q4cOpX379kRFRTFw4ECOHTtGUFAQc+bMoV27dqhUKoyNjUvt51OUsgP+1q1bc+7cOd555x28vLwICAjgwYMHLF26lCZNmjBlyhTCw8NJT0/H398fDw8PgoOD2bt3LyqVCrVazbZt20r7dIpcp06d+Oabb+jXrx9mZmbExMRgZGREjRo1tHXu3LnDm2++ibe3N2lpaYSFhUkAKoQQosKQAFSIMq5SpUraxwYGBrmmdWYHLQYGBtpppBqNhhUrVtCwYcNc7Vy4cEE7CvO8evfv339hnQsXLuTql76+fpmeblq7dm3atWsHQL9+/Vi1ahXwT7Kn8+fPc+XKFXr37g1Aeno6bdq00R7fp08f7b//7//9PwBiYmIYOXIk9+7dIy0tjXr16mnru7u7Y2pqCsCJEye4cuWK9rWkpCRUKhXt2rVj+vTp9O3bl27dumlvMlQEN2/eZPny5SxcuJD333+fHTt2sGPHDg4cOMC3335Lo0aNCjRKXxG5uLhw9epV7f+9ypUr8+233+YKQE+dOkVQUBCGhoaYmZmxePHi0uquEEKIV3DgwAGuXLnCqFGjCAgIwMzMjBEjRuism/27f+3atSXcy5InAagQFYyLiwurV69m1qxZ6OnpcfnyZVq0aPFS9QraVlmnp6en83nOaZHOzs58//33Lzw++/FXX33FsGHDcHd359SpUyxcuFBbJ2egr1arCQ0NxcTEJFebo0aNwtXVlcOHD9OnTx9+/vnnPIF+eWVjY0OzZs2ArNFxJycn9PT0aNq0KZGRkcTExBRolL6isLGx4fDhw9rnQ4YMYciQIXnqXb16FQAvLy+8vLxKrH9CCCGKh7u7O+7u7qXdjTJHkhAJUcH4+vqSnp6OQqGgc+fOzJ8//6XrFbStsi4qKopz584BsGPHDu1oaLY2bdrw+++/a5PBJCcnc/36de3r2VmJQ0JCtCOjjx8/xtraGoAtW7bk+97ZQXy2y5cvA1mjhM2aNePzzz/H3t5em0SqIsg5nVhfX187Wq6vr09mZqZ2ZF2pVKJUKvn999+1ibdyBu+vG/20h1SJ3UTVOz9QJXYT+mkPS7tLQrBy5UpcXFwYNWpUgY/x8fEhISGBhIQEfvrpp+LrnBClKDIyEmdnZ3x9fXFycmLUqFEcP36c3r174+joyB9//EFwcDCTJ0/Oc2xERAQDBgxAoVDg4eHBzZs3gazvH7oST1Y0MgIqRBmUPRLy3nvv8d5772nLZ8+erX2cc29Ae3t7tm7dCoCpqanOQHHAgAEMGDBA+zy/ejnfU1cdVVImjer1oEEdAy6cVtGkpUmZny5ia2vLmjVr8Pf3p3HjxgwaNChXUFi9enUCAwP5/PPPSUtLA2D8+PHY2toCkJCQgEKhoFKlSnz33XcA+Pv7M3z4cCwtLXF0dCQyMlLne8+cOZNJkyahUCjIyMjAwcGBefPm8eOPP3Lq1Cn09fVp3LgxnTt3LuafQtlRUUbWi5J+2kOqRq/CMCNOW2aUGsmj2oNRV6peij0Tr7s1a9awadOmXMsEXpQkbN26dUDWF/S1a9fyySefFHc3hSgVL1py4uHhofO40aNH8/nnn9OtWzdSU1PRaDRER0c/N/FkRSIBqBCiwFRJmZw5mkSySgOkAxD/MIMOncwxMzco3c49h6GhId9++22uspwBPICTkxN79uzRefzIkSPz3MEJ24ezAAAgAElEQVT08PDQ+YfF398/13MrKyvtVNOcyvvWNa/C19eXadOmoVAoUKvV2NjYlPmbGMXNLE6ZK/gEMMyIwyxOSaL1v0upV+J1N2HCBG7fvo2Pjw9RUVG4ublpM3y7uLhw6dIl7Y3Rjz/+mMjISI4cOYKDgwN79+5lzpw53Lp1Czc3N5ydnRk2bBgjR44kMTGRzMxM5s6di4ODQymfpRAv70VLTnRJSkoiJiZGm0gy5xKd5yWerEgkABVCFFh4WOrT4PMfySoN4WGptH7XrJR6Vb6okjIJD0slNUWNiak+TVqalGrw3qhRI+Lj44usvWfXOy5atEjnawUZpX+dGGQkFqpciJIwb948jh49ypYtW1i9ejVKpVKb4Ts4ODhP/ZyzdAAmTZpEeHg4SqUSgKCgIFxcXBgzZgyZmZmkpKSUyHkIUVxetOSksJ6XeLIikTWgQogCS01R6y5P1V1eFjwbEBXW2bNnsbKyKpK+ZI8gR91O5+H9TKJup3PmaBKqpML/kXpWef0jpb4fi/rHADIXTEb9YwDq+7Gl3aVSkWlYpVDlQpSGnBm+dfnoo484deoUDx480JbdvXtXG6xevnyZwMBA7O3t8fPzk72PxWvJ3NycN998k3379gFZifhet5sxMgIqhCgwE1N9IG+wZGIi97IKoiAjyIGBgWzfvp3q1atTu3Zt7OzsqFKlChs2bCAtLY233nqLJUuWYGpqiq+vL8bGxvz555+0bduWR48eUaVKFS5evMj9+/eZPHkyPXr0AGDZsmWEhoaSlpZG165dGTduXJ7+6aoTGRnJRx99RPv27Tl37hzW1tasWrXquV9CC0p9PxZN4FR4GnRqAG6Eo/abwf9n784Dmjzyx4+/Ew5BEAVBqRbdoqJWRa2lqNjiEfAARS1ilVXb9Ww9viotSvFY16OixdtW7aqr1p9FvBG1Bi+0HutR8WgXL8CIYBUQJFwJye8PSgoSPJFwzOsvMpnneebJEMhkZj4fqZ39a5+/MlHaeGCSoyi2DFdtbIPSxsOArRKE4ooGCTM2Nkaj+evLx9zc3Gcem5qaSkxMDGfOnOHo0aOsX7+e8PBwkeNWqJZWrFjBtGnT+PbbbzE2Nmbt2rWGblK5qlafGhUKBd27d3/h+ocOHSqWv08QqrvmbcyoaVE8pUlNCwnN25iVcoRQ1PNmkC9fvsyBAweQy+X8+OOPxMTEANC7d28OHDhAVFQUTZs2Zdu2bbpjk5KS2Lt3ry4/6YMHD9izZw+bNm3im2++AeDEiRPExcURGRnJ4cOHuXLlCmfPni3WhmfViYuLY8SIERw7dgwrK6tS98q+tL1bdYNPnYfJBeXVjMa0Lo8b/INsy7bkmTmSbdlWBCASKjQHBweuX7+ORqMhMTGRy5cvl6hjYWGBSlUQL8DKygqpVEpISAjW1tYMGTKEq1evlnezBaHM6NtyUvilb+FzgwcP1i1NDwgI0OUAdXR0JDw8nKioKA4dOkTjxo3p3LlzsXgI8+fPr7LbUsQM6DMcOnQImUyGk5OToZsiCBWChaURHbtaEns1h/x8I4yM8g2+h7Eyed4M8vnz5+nZs6cuIIGHR8HsV2xsLIsWLSIjIwOlUom7u7vuWG9vb4yM/nr9e/XqpYus+/DhQ6BgcHnixAldLrKsrCzi4uLo2LGj7rjS6jRs2BAHBwddlFpnZ+dSAyu8LO3j1Jcqr+o0pnVFwCGh0nBxcaFRo0Z07dqVZs2a0aZNGy5dulQsOq6NjQ22trYsWrSIGzduMHbsWJYtW4ZcLic/P5/Dhw8b8A4EoWJJzsxja8wj0rJUWNc0wb+tLfaWps8/sBIy2ABU3zKzXr16ERwcTEpKCubm5ixevJimTZsyefJk6tSpw4ULF0hJSSE0NJQdO3Zw8eJF2rdvrwty0axZM136iv379/Pdd9+Rk5ODWq3WbeqNj49nyJAh3L9/n7p165KRkYFEIuGPP/6gdu3amJqaYmVlRUpKCnfu3OHnn39m+fLl/PDDD/ztb38z1MslCBWGhaUR73WywNbWttg+H+H5mrcxIy1FXWwZ7ovMIE+ZMoX169fTqlUrwsLCOHPmzF/HP5U3s2gAg8L8YVqtlgkTJjBs2LBSr1FaHYVCUSzIgpGRETk5Oc9s74uS1LFBX4YzSZ2y2XMrCMLrK4wY/nSEb4lEwqpVq4qVNWvWjIYNG2Jra4uFhQXp6eloNBoCAwPx9vYmOzuboUOHkpGRQadOnWjUqFG53YcgVGTJmXnMPqIgOVP1Z0k2Nx5lM6eHQ5UchBpkCe758+f1LjMLDAxk7ty5HDp0iJkzZxIUFKQ75vHjx0RERPDPf/6Tzz77jNGjR3Ps2DH+97//6RK7F5WQkEBcXBz79++nZcuWZGZmsmzZMnJzc7lz5w4nTpxAoVDg6upKWFgYlpaWHDlyhJkzZ9KoUSOOHz/OwIEDady4MXPnzhWDT0EQXlvhDHLDRibUrWdEw0YmxVLYuLi4IJfLycnJQalUEhUVBRSEbK9fvz4qlYrdu3e/9HW7du1KWFgYSqUSKFi2+/SXBy9Sp8z5+MPTez3t7AvKBUGodCQSCQ0bNqRv3750796dcePG0bp1a6RqNdL/xTJy8Cd4fvQRA3x8mD17tqGbKwgVxtaYR0UGnwWSM1VsjamaX/QbZAb0l19+KbHMLCcnh4sXLzJ27FhdvcKE8ABeXl66vDq2trbFcu7cu3evRBLzW7du0bBhQ2rWrMmZM2eQSCSMGDECY2Nj1Gq1bhnb7t27+dvf/oaxsTHTpk0jLS2N69ev4+npSWJiIqampiWWqgmCILyqwhlkfdq1a4enpycymQw7OztatmxJrVq1+Oqrr/D29qZu3bq0b9+ezMzMl7qmu7s7N2/epF+/fkDBrOnKlSuxtbV9bp2iy3vLmtTOHs2Uf8HerWgfpxbMfPr4V7sARELpJk+ejEwm0+2rqmznr05SU1OpU6cOADNmzGDGjBkAGOXmYnM7DpM8FedDlwKgMjUhtck7ejYkCEL1lJal0l+erb+8sqswe0C1Wi1WVla6XFFPK1wCJpVKS+TcKUw/IJH8FRylaEoCjUbD0KFDqV+/PmvXruW///0vAP379+fu3bvk5eWRlZVF+/bt2bBhAzVq1EAul4t/TIIglLtx48YREBBAdnY2AwcOxNnZmTZt2jBixIgSdYvm2NT3uHBLAsCoUaMYNWpUiXO8SJ2iQRYKAyiUFamdPYwKeH5FQRAqrOTkZHx9ffX+fTBWJGGSV/xDtEmeilrJD3jcWCzBFQQA65omQMlULNbmJuXfmHJgkCW4bm5uJZaZmZub4+DgQEREBFAwIL1+/fpLndfOzo6bN2+i0Wh48OAB9+/fJzs7Gzc3N7Zv346rqyuJiYls374dgC1bttC9e3c+/fRT3n33Xezs7AgMDCQhIQGlUomlpSWJiYlin5tQbaxfvx53d3cmTJjwQvUVCkWxJaExMTHMnDnzTTVP5/DhwyX2Hj3drtIiXoeFhZGcXHFzTQYGBuLh4UHPnj3p06cPbdq0MWh70tPT+fnnn9m5cyc///wz6enpBm2PUPktXbqUDz/8kP79+/PFF1+wZs0a4uPj8ff3p1evXgwYMIBbt27p6p88eZLevXvTpUsX3ZfUCoWCAQMG0LNnT3r27Mn58+d19b///nv69OmDTCbj22+/1ZWHh4cjk8mQyWRMnDixRLsWLVrE5MmTXyl5fFX2Iv8X7O3tOXXqFP/4xz8AGDZsGOnp6dy8/4jQH7fpPcYor3LmLhaEN6Eg4FDxwaa9ZUEgoqrIIDOgLi4uepeZrVq1iqCgIJYvX45arcbHx4dWrVq98HmDgoIYMWIENjY2tG3blszMTLy8vFCr1dSrV48vvvgCExMTVqxYwerVq1GpVMTGxrJt2zYePXqEQqFAIpHQt29f+vXrR1ZWFn/88Qfbtm3jP//5j9gHKlR5mzZt4qeffqJBgwbPratWq3UD0AEDBgDQtm1b2rZt+0bbqFar8fT01EVrfVnh4eG0aNECe/uKucxz9erVhm6CTnp6Ort37yYjI0NXlpSUxIABA6hdu7YBWyZUVkVTDanVanr27ImzszOBgYEsXLgQR0dHLl26RFBQEOHh4QDcu3ePyMhI4uPjGTRoEB9++CG2trZs27YNMzMz7ty5w/jx4zl48GCxdEJarZZPP/2Us2fPYm1tzfLly9m3bx82NjakpaUVa9fcuXPJzMxk6dKlxVZTCfr/L6jV6mLRbp+2ZcsWAEKPXiDqYAQL+pdcSZZvWmEW4QmCwdlbmjKnh0NBFNxsFdbmIgruG6FvmVmjRo3YurVk/rdly5bpIm7qy7lTyNvbWyyXFYRXNG3aNO7evcuwYcPo168fCQkJxMbGolKpCAgIoGfPnoSFhXHw4EGUSiVGRkZkZmZy69YtPDw8GDRoEK1bt2bNmjVs3ryZlJQUxo8fz4MHD+jQoQPR0dEcOnQIGxsbvVGwx40bR3x8fKmRsGvUqMH169d5//33admyJVeuXGH+/Pk8fPiQ6dOnk5CQAMA333yDvb09+fn5fPXVV1y4cAF7e3s2bNjAkSNHiImJYcKECZiZmbFv3z5u3rzJnDlzUCqVurbVr1/fwL1RMZw9e7bY4BMgIyODs2fP0rNnTwO1SqjM9KUael4MiL59+yKVSnF0dKRx48bcunWLRo0aERwczG+//YZUKuXOnTtA6emEfvvtN7y9vbGxKYiwbG1trTv/smXLeO+991i0aNEbv//Kpuj/hcTERDw8PLh79y4NGzbE3d1d93cYYPjw4YwbN47OnTvj6urKwYMHObp5BYoHSbQa/wV93mvP4pGjAXigloC9+DsrCEXZW5oS4Pb8CYCqwGAD0MDAQG7cuEFubi6DBg0y+DKzQslJaRw7cprsHCXmZhZ069EZ+7esn3+gIFRyISEhHD9+nPDwcNauXYubmxtLliwhPT0dLy8vPvzwQwCuXr1KVFQUzZo1Y9++fboBJ8Dp06d151uyZAlubm5MnDiRY8eOsW1bwTKs0mZAgGfOgiQlJbF3716MjIwICwvTXWfmzJl07NiR9evXk5+fj1KpJD09nbi4OFavXs3ixYsZO3YsBw4c4OOPP+Y///kPM2fOpG3btqhUKmbMmMHGjRupW7cue/fuJSQkhCVLlpTLa17RlRbsqDBSriCUhefFgHh6RlIikfDDDz9gZ2eHXC5Ho9Hg6OioO5e+dEIbNmwo9frt2rXjypUrpKWlFRuYCsX/L2zcuBG5XM7u3bsxNzcv9ne4NN2HT2Jrwm2a/N9autfTcD5TyyO1hEumVnxWJJ6HIAjVi0H2gELBMjO5XE50dLTevRiGkJyUxu7de3iYepvMrGQept5m9+49JCelPf9gQahCoqOjWb16NR4eHvj6+pKbm0tiYiIAH3300Qt9SPvvf/+Lj48PAN26ddNFRyw6A2JpaYmHhwdQMKgpnAXx8PBg2rRp/PHHH7rzeXt7643I+ssvvzB8+HCgIEellZUVAA4ODrro2M7OzigUihLH3r59m9jYWD755BM8PDxYsWIFSUlJL/w6VXWWlpZ6yy0s9EfxFYTn0Zdq6HkxIPbv349GoyE+Pp6EhASaNGlCRkYG9erVQyqVsnPnTt2+zdLSCbm5ubF//35SU1MBii3B7dq1K+PHj2f48OEvHWG6uvH09MTc3PyF6/dvaYOJVEKSSsqsRGO+SDDhu3RzejtXzC0QgiCUD7EAv4hjR06jUj8pVqZSP+HYkdMM+buXgVolCOVPq9Wybt06mjZtWqz80qVL1KxZ841cU6PRPHMW5GWvWzRatpGRETk5OSXqaLVanJycdB98heI6duxIUlJSsWW4VlZWIi2V8MpKSzX0rBgQDRo0wMvLiydPnrBw4ULMzMwYMWIEY8aMYceOHXTr1k3396G0dELNmzdn0qRJ+Pr6IpVKad26dbEtPH379kWpVPLpp5+yZcuWlxpkVSdF/w4bGxuj0Wh0j3Nzc0vUt7MwoZ6lCR/9zapa7GsTBOHFiAFoEdk5+peVZeeK5WZC9eLu7s7GjRuZN28eEomEa9eulci1CwUzZKUtx3RxcSEiIoLx48dz4sQJHj9+rCufNm0aEyZMID8/n6ioKP7+979Tq1Yt3SxI37590Wq1/Pbbb88NRNalSxc2b97M6NGjdUtwn8XCwkI3y9GkSRNSU1O5cOEC77//PiqVijt37tC8efMXeZmqvNq1azNgwADOnj2LUqnEwsKCjh07igBEwmt52RgQ+jg6OhIVFaV7HBwcrPu5tHRCfn5++Pn5lXr+Tz75hE8++eSl76e6cnBwYNOmTWg0GpKSkrh8+XKJOhYWFuRkKavNvjZDunXrFp9//jkSiYR169aVGjizcH9u4X5oQTAEMQAtwtzMgswsPeU1xHIzoXqZPHkys2fPRiaTodFocHBw0O3zLKply5ZIpVJkMhl+fn7FBqlTp07liy++YOfOnXTo0IF69ephYWFR6gwI8EqRsP/1r38RGBjITz/9hFQq5ZtvvnlmECE/Pz+mT5+uC0K0du1aZs2aRUZGBvn5+YwaNUoMQIuoXbu2CDgklKmKEANC8zAZ9m5F+zgVSR0b8PEvyEkrvDAXFxcaNWpE165dadasmd5+tLGxwcXFhe7du9OtW7dySdNVXR06dAgvLy8mT55s6KYIwnNJtFqtIa6rvX///ksdUBgF900q3ANadBmuiXEtBgzoLwIRlZHy6EehfDyvL3NzczEyMsLY2JgLFy4QFBSkW15bOJtWOAOyaNGiChOIrDoS78uqQfTji9E8TEa7dBY8LJIP2M4eyZR/VZhBaGXuy/T0dM6ePUtmZiaWlpbVetXE6/bjP/7xD+7fv09ubi4jR47k73//O82aNWPo0KFER0djZ2fH999/z+XLlwkICMDIyIh33nmHHTt2sHPnTjZs2EBeXh7t27fnm2++wcjISMyAvqLK/J40pD/TN5XIbSVmQIuwf8uaAQP6F0TBzVViXkNEwRWEV5WYmMi4cePQaDSYmpqyePFi3XMVYQYEIDkzryDnVpYK65pib5IgVAt7txYffELB471bYVSAYdpURYjcwWUrNDQUa2trsrOz8fLyok+fPmRlZdG2bVvmzJnD0qVLWbJkCfPnz2fYsGFYWFgwbtw4bt68yb59+9izZw8mJiYEBQWxa9cuBg0aZOhbEgRADEBLsH/LWgQcEoQy4OjoyOHDh/U+t3r16nJuTUnJmXnMPqIgOVP1Z0k2Nx5lM6eHgxiECkIVpn2c+lLlwosTuYPL1oYNGzh48CAA9+/fJy4uDqlUqguyNXDgQL37nU+dOsXVq1fp06cPADk5Odja2pZfwwXhOcQAVBCEamlrzKMig88CyZkqtsY8EgEzBKEKk9SxQd/mI0kdsSTxdYncwWXn9OnTnDx5koiICMzNzXUp0Z72dJ5cKIjwPmjQIIKCgsqjqYLw0gyWB1QQBMGQ0rJU+suz9ZcLglBF+PjD03s97ewLyoXXInIHl50nT55Qu3ZtzM3NuXXrFpcuXQIKUpZFRkYCsHv3bj744IMSx3bp0oX9+/fr9iympaVx79698mu8IDyHGIAKglAtWdc00V9urr9cEISqQfpnwCGJqzs0b4PE1b1CBSCqzDp27IiVlVWxMpE7+NV07dqV/Px83N3dWbBgAe+99x5QkIv1119/pXv37vzyyy9MmTIFKNh/u2bNGgCcnJwIDAxkyJAhyGQyhgwZwoMHDwx2L4LwNBEFVyhXoh+rjsrelyX3gIK9pUm13ANa2fuyqjl06BCOjo44OTm91HGrV6+mTZs2fPTRR8XKT58+zZo1a/SmUhIqpsr8niyMgityB7+ZfmzWrBk3b94sUa5QKBgxYgRHjx4t0+sJBSrze9KQRBRcQRCEIuwtTZnTw6EgCm62CmtzEQVXqBgOHTqETCbTOwBVq9UYG+v/1z179mzxAUkwOJE72HDUajUTJkzg6tWrODk5sWLFCtauWsWRQz+Tk5ODq7MzC5YuQWNmhq+vL+3bt+f06dOkp6cTGhqKq6srCoWCSZMmkZWVBcC8efNwcXHh9OnTLFmyBGtra2JjY3F2dmblypVIJBKuXLnCnDlzUCqV2NjYsHTp0mfm4xYEMQAVBKHasrc0FQGHhHKxdOlSdu3aRd26dWnQoAHOzs706tWL4OBgUlJSMDc3Z/HixaSlpSGXyzl79izLly/nhx9+4Msvv+Tdd9/l/Pnz+Pj40LlzZ70f9kaNGkWXLl3w9vbm2LFjzJ49G3Nz82J7xLKyspgxYwaxsbGoVCoCAgLEYEEQKhF9s5+Fbt++TWhoKC4uLkydOpXN69cT8KE7izx6ATDs20Wc2fr/6Pz3gv3OarWayMhIjhw5wpIlSwgLC8PW1pZt27ZhZmbGnTt3GD9+vC4S77Vr1zh69Cj29vb4+Phw/vx52rdvz4wZM9i4cSN169Zl7969hISEsGTJkjf/YgiVlhiACoIgCMIbdPnyZQ4cOIBcLketVtOzZ0+cnZ0JDAxk4cKFODo6cunSJYKCgggPD8fDwwOZTIa3t7fuHCqVioMHD6JSqfj444+f+WEvJyeHr776iu3bt/POO+8wbtw43XPLly/Hzc2NJUuWkJ6ejpeXFx9++CE1a9Ys19dEEISy16BBA1xcXICCFC1bVq3mlMSIRTvCycrNJTXzCa0aNaZncsF+0MI0Lc7OzrogRSqViuDgYH777TekUil37tzRnb9du3aFSypp1aoVCoUCKysrYmNj+eSTT4CCIEn16tUrt3sWKicxABUEQRCEN+j8+fP07NkTMzMzADw8PMjJyeHixYuMHTtWVy8vL6/UcxTm/bt9+/ZzP+zdunWLRo0a4ejoCMDHH3/Mjz/+CEB0dDRyuVwXrCQ3N5fExESaNWtWRncrCIKhPJ2SRarV8sV3q7iwbCUOdnb8c+sWclR5GOWpATA1LdhyYmRkhFpdUPbDDz9gZ2eHXC5Ho9Ho/o4UrV/0GK1Wi5OTExEREW/69oQqRAxABUEQBKGcabVarKyskMvlL1S/cIbydT/sabVa1q1bR9OmTV/peEEQKq7ExEQuXLjA+++/z549e+jYrh1nr13F1sqKzOxsdvxyCl+3LuSblv7xPyMjg7feegupVEp4eDj5+fnPvGaTJk1ITU3VXVelUnHnzh2aN29e1rcnVCEiDYsgCIIgvEEuLi7I5XJycnJQKpVERUVhbm6Og4ODbiCp1Wq5fv06UJBLUalU6j1X0Q97ULBcLjY2tlidpk2bolAoiI+PB2DPnj2659zd3dm4cSOFEfCvXbtWpvf6ohQKBd27dy9R/uWXX3Ljxo0yu07R861YsaLMzisIFVGTJk3YtGkT7u7upKen88nn4/hHHy9afzGOnjODcWnmRL6REU/sSw8QNGLECHbs2IFMJuPWrVvPXZ5vamrK2rVrWbBgATKZDE9PT93fJ0EojUjDIpQr0Y9Vh+jLqkP05ZsXGhrK7t27sbOzw9bWlq5du/Lhhx8SFBTEgwcPUKvV+Pj4MGXKFM6fP89XX32Fqakp69at48svv2TmzJm0bdsWKBg0zpo1i4yMDPLz8xk1ahT+/v5Mnz5dbxAiV1dX4uPj2bx5M9nZ2cyePZuLFy+i0WhwcHAwSHoWQ6SMKC19RUUk3pNVQ0XoR6PcXGolP8AoT02+qTFP7OuTX6OGQdtUGVWEvqyMSkvDIgagQrkS/Vh1iL6sOgzdl2FhYVy5coX58+eXeM7V1ZWDBw9iY2NjgJaVncKciNnZ2QwcOJBFixbRpk2bMr2GofvxZSgUCvz9/XF2di6WMmLYsGHMnDmT1q1bExAQwJUrV5BIJAwePJgxY8YQFxfH9OnTSUlJwcjIiLVr13L//v1ieU6Dg4NxdnZm8ODB+Pr6MnPmTCIjI/n+++9p0aIFzZs3Z9GiRYwdO5akpCQ0Gg3/93//h4+Pj4Fflb9Upr4USlcd+jE5M68gnVmWCuuaVTedWXXoyzdB5AEVBEEQBAMJDAzkxo0b5ObmMmjQoDIffL6IzNw/uPpwB9mqx5ib1KGNnS+WNQwXrfLplBGbNm3SPXf9+nWSk5N1M6Tp6ekATJw4kfHjx9O7d29ycnLQarW8yBfaX3/9NRs3btTtuY2MjMTe3p4tW7YABfveBEF4OcmZecw+oiA5U/VnSTY3HmUzp4dDlRyECmVH7AEVBEEQXptCoeCjjz5i8uTJdOnShQkTJhAdHY2Pjw9ubm78+uuvZGVlMXXqVLy8vPD09OTnn3/WHf/gwQP8/f1xc3Nj3rx5BryTN2P16tXI5XKio6OZOHFiuV8/M/cPjieEcDf9DA+zfudu+hmOJ4SQmftHubel0NMpI/773//qnmvUqBF3795lxowZHDt2jFq1apGZmUlSUhK9e/cGwMzMDHNz81e6dosWLYiOjmb+/PmcO3cOKyur178hQahmtsY8KjL4LJCcqWJrjJgpFJ5NDEAFQRCEMhEfH8/YsWOJjo7m1q1b7Nmzhz179jBr1ixWrlypy0EZGRlJeHg4c+fOJSsrCyiY8fr+++85cuQI+/btIzEx0cB3U7VcfbgDpar4YFOpKpgRNZSnU0YUfVynTh3kcjmdOnViy5YtfPnll6Wex9jYmKLbiXJzc5977SZNmnDo0CFatGjBokWLWLp06SvcgSBUb2lZKv3l2frLBaGQGIAKgiAIZcLBwYGWLVsilUpxcnKiS5cuSCQSWrRogUKhIDo6mtWrV+Ph4YGvr68uByVAly5dsLKywszMDCcnJzEALWPZqsd6y3NKKS8PhSkjoCBSb+FsKEBqaioajQYvLy8CAwO5evUqlpaWvPXWWxw6dAgoGGhmZ2fTsGFD3fLm9PR0Tp06pfd6JiYmqFQFH4yTk5MxNzfn448/Zty4cVy9evUN360gVD3WNU30l5vrLxeEQmIPqCAIglAmahSJrLpPE3QAACAASURBVCiVSnVJy6VSKfn5+RgZGenNQXnp0qViCc6lUqkuKbpQNsxN6ugtNyulvDwUpowICAjAycmJESNGEBUVBUBSUhJTp05Fo9EAEBQUBBSkUpk2bRrffvstxsbGrF27lsaNG9O3b1+6d+9Oo0aNaN26td7r+fv7I5PJaNOmDb6+vsybNw+JRIKJiQnffPNN+dy0IFQh/m1tufEou9gyXHvLgkBEgvAsYgAqCIIglIvCHJSFH/yvXbtW6mBBKFtt7HxJybpdbBmuhUk92tj5GqQ9Dg4OREdHlyjfseOvJcFF9wgXcnR0JDw8vET5jBkzmDFjxjPPFxwcTHBwsO5x165dX7bZgiAUYW9pypweDgVRcLNVWJtX3Si4QtkSA1BBEAShXEyePJnZs2cjk8kMmoOyOrKsUY+ujadx9eEOclSPMasAUXANobqkjBCE8mJvaUqAWwNDN0OoZEQeUKFciX6sOkRfVh2iL6sG0Y/PVjJlRMFywYqYMkL0ZdUg+rHqEH35akrLAyqCEAmCIAgVhlFuLnUS7lL35h3qJNzF6AUimgqGc/r0aYYPH27oZrwQkTKicilM7TRhwgTc3d0ZPXo02dnZnDx5Ek9PT3r06MHUqVN1UY9dXV355ptv8PDwoHfv3ly9epWhQ4fSuXNn3UqL06dP8/HHH/PZZ5/RqVMnFixYwK5du/Dy8qJHjx7Ex8cDkJKSwujRo+nTpw99+vTh/PnzhnoZBKFKEgNQQRAEoUIwys3F5nYcNdPSqaFUUjMtHZvbcWIQWk6qeuAnkTKi8rl9+zYjRozgxIkT1KpVi7Vr1zJlyhRdyia1Wl1sGX+DBg2Qy+V88MEHTJkyhXXr1hEREUFoaKiuzm+//cbChQs5fvw4O3fu5M6dO0RGRjJkyBA2bNgAwKxZsxg9ejQHDhzghx9+eGYaIEEQXp4YgAqCUGn069ev3K/p6upKamrqa18/LCyM5OTksmpWlVQr+QEmecUHAyZ5KmolPzBQiyqv8PBwZDIZMpmMiRMnljqjExoaysSJE/Hx8WHSpEml1vv111/p27cvnp6e9OvXj1u3bpW4ZlZWFlOnTsXLywtPT09dEKHY2Fi8vLzw8PBAJpNx586d8nshihApIyqfBg0a6NLzDBw4kFOnTtGoUSOaNGkCwKBBgzh37pyuvqenJwAtW7akffv2WFpaUrduXUxNTUlPTwegbdu21K9fnxo1atC4cWPc3d0BaNGiBffu3QPg5MmTBAcH4+HhwaeffkpmZiZKpbLc7lsQqjoRhEgQhEpj3759ZXIetVqNsfHL//l7neuHh4fTokUL7O3tX/kcVZ1Rnv4ZuNLKBf1iY2NZvnw5+/btw8bGhrS0NGbMmMHo0aP54IMPSExMZOjQoZw4cQKAmzdvsnv3bszNzRk/frzeek2bNmX37t0YGxsTHR1NSEgIP/zwQ7HrLl++HDc3N5YsWUJ6ejpeXl58+OGHbNmyhZEjRzJw4EDy8vLIz883xMsiUkZUQhJJ8a1jtWvXJi0trdT6hamgJBJJidROhb93T5cXTRdVuApAo9EQERGBmZlZ2dyIIAjFiAGoIAiVRrNmzbh58yanT59myZIlWFtbExsbi7OzMytXrkQikXDkyBHmzJlDzZo1cXFxISEhgc2bNxMaGkp8fDx3796lYcOGzJ07l+nTp5OYmAjAnDlzcHFxITU1lfHjx5OcnEyHDh0oGqit8PoA33//PREREeTl5dGrVy++/PJLFAoFf//73/nggw+4cOEC9vb2bNiwgSNHjhATE8OECRMwMzNj3759LF26lMOHD2NsbMxHH33ErFmzDPKaViT5psagZ5Ih31T8q3oZv/zyC97e3tjY2ABgbW3NyZMnuXHjhq5O0RkdT09PzM3NAUqtl5GRweTJk4mLi0MikaBSlVy2Gh0djVwuZ82aNQDk5uaSmJhIhw4dWLFiBUlJSfTu3RtHR8c3du/PIlJGVD6JiYlcuHCB999/nz179uDs7MyPP/5IXFwc77zzDjt37qRjx45lft3ClFGff/45gEgZJQhlTPxXFwShUrp27RpHjx7F3t4eHx8fzp8/j7OzM9OmTWPXrl00atSIL774otgxLzLTs3TpUt3+oaioKLZt21bi2idOnCAuLo7IyEi0Wi2ffvopZ8+epWHDhsTFxbF69WoWL17M2LFjOXDgAB9//DH/+c9/mDlzJm3btiU1NZWDBw8SHR2NRCLRLQ2r7p7Y18dEmVVsGa7K1IQn9vUN2Kqq4VkzOjVr1nxuveDgYDp37sz69etRKBT4+pbMH6rValm3bh1NmzYtVt6sWTPat2/PkSNHGDZsGCEhIXTp0qWM7uzliJQRlUuTJk3YtGkTAQEBODk5MXfuXN577z3Gjh1Lfn4+bdu2ZdiwYWV+3blz5/L1118jk8lQq9W4uroSEhJS5tcRhOpKDEAFQaiU2rVrVxjem1atWqFQKKhZsyaNGzemUaNGAPTv358ff/xRd8yLzPScPXuWf//73wDIZDLq1KlT4tonTpzgxIkTuv1GWVlZxMXF0bBhQxwcHHTflDs7O6NQKEocb2VlRY0aNQgICNDt0xMgv0YNUpu8Q63kBxjlqck3NeaJfX3y/1xWJ7wYNzc3Ro4cyZgxY3RLcF90Rqe0ek+ePNEtH9++fbve6xYeO2/ePCQSie7YhIQEGjduzMiRI0lMTOT333832ABUqFyMjY1ZuXJlsbIPP/yQw4cPl6hbdC/o4MGDGTx4cInnOnfuTOfOnXXlO3bs0P1c9DkbGxvdTL4gCGVPDEAFQaiUiu7jMTIyeqEIni8y0/MitFotEyZMKPHNu0Kh0O1BKmxXTk5OieONjY2JjIzk1KlTREZGsnHjRsLDw1+6HVVRfo0aPG7cyNDNqNSaN2/OpEmT8PX1RSqV0rp16xee0Smt3ueff87kyZNZvnw5PXr00HvdyZMnM3v2bGQyGRqNBgcHBzZv3kxERAQ7d+7E2NiYevXqMXHixDf9EgjCK9M8TIa9W9E+TkVSxwZ8/JHaib37glCWxABUEIQqo0mTJiQkJKBQKHBwcHhm0KDSZno6duzI7t27mTx5MkePHuXx48clju3atSuLFy9m4MCBWFhYkJSUhInJsyNpWlhYkJmZCYBSqSQ7O5sePXrg4uJCp06dXuOuBaEkPz8//Pz8ipXpm9EJCAgo9ri0mZ/333+fU6dO6R5PmzYNKD5rZG5uzqJFi0ocO2HCBCZMmPDyN1EJFd0nXpp+/fqxb98+FAoFFy5cYMCAAeXUusrFwcGBo0ePlus1NQ+T0S6dBQ8LIpZrAe7EopnyLzEIFYQyJNKwCIJQZZibm7NgwQL8/f3p1asXFhYWWFlZ6a07d+5cYmJikMlkdO3alS1btgAwZcoUzp07R7du3Th48CANGzYscay7uzv9+/enX79+9OjRgzFjxugGl6Xx8/Nj+vTpeHh4kJmZyYgRI5DJZAwYMIDZs2e//s0LwgsIDQ195tLCzZs3l+lsvFFuLnUS7lL35h3qJNwVOV35K5q2QqFg9+7dBm6NYSkUCrp3727oZvxl71bd4FPnzxlRfcLCwggODgYKVgDs37//TbdQEKoEMQMqCEKlUTiz8PQ+nvnz5+t+dnNzIzo6Gq1Wy9dff42zszPw4jM9NjY2egMPFb0+wKhRoxg1alSJOkW/sR83bpzuZy8vL7y8vHSPIyMj9d+kIBjQ8OHDy+xcRrm52NyO+yuolBJMlFmkNnlH777eV02PVBEplUo+++wz0tPTUavVBAYG0rNnT+CvWdIFCxZw69YtPDw8GDRoEGPGjDFwqwXt49Rnlufn52NkZFR+7dFq0Wq1SKVivkioWqrGX3pBEIQ/bd26lfDwcFQqFa1bt34jERJfhzIzn9irOeRkazAzl9K8jRkWluX3gUaofpYvX054eDi2trY0aNAAZ2dn4uPjCQ4OJiUlBXNzcxYvXkzTpk0JDQ3FwsKCcePGlVpn8uTJyGQyvL29gdLTI3Vo0oSfJk0BiYSLN28y9d/ryMzOxrpuXRav+Z769evj6+vLu+++y/nz5/Hx8Sn2pU1lVqNGDdavX0+tWrVITU2lb9++eHp6Fstr+fXXX7NmzRo2b95swJYanlqtZsKECVy9ehUnJydWrFjBmjVrkMvl5OTk8P777xMSEoJEIsHX15f27dtz+vRp0tPTCQ0NxdXVFYVCwaRJk8jKygJg3rx5uLi4PDNl15UrV5gzZw5KpRIbGxuWLl3KyYRkNvzyOyqNlnZ1LJjfuhGtDl/GPw1OyWQsWLCAuLg4Vq5cSe3atXn33XeLxSM4efIkq1ev5smTJ8yePZshQ4aQn5/PggULOHPmDHl5eYwYMUL3f6m0dF5Dhw6lffv2XL16lS1bthAREVGiniBUZmIAKghClTJmzJgKO5OgzMzn7PFMspSFuUXzSUtR07GrpRiECm/ElStX2LdvH3K5HLVaTa9evXB2diYwMJCFCxfi6OjIpUuXCAoKKrH09kXqPK1oeiTfnr345bfruDZvwcQ137F31mzsatfhx3NnCQkJYcmSJQCoVCoOHjz4xl4DQ9BqtSxcuJBz584hkUhITk7m4cOH1KtXz9BNqzDWr1/P+vXrSUhIIDQ0FBcXF6ZOncqmTZv49NNPmTJlCgATJ05ELpfj7OxMbGwsbdq0ITIykiNHjrBkyRLCwsKwtbVl27ZtmJmZcefOHcaPH6/7ndKXsqt9+/bMmDGDjRs3UrduXfbu3UtwcDAqpZJd/dwxSfmD4Gt32Z2YSla+hvb9ffnn34fz4MEDxo8fz6FDh6hVqxaDBg0qFk363r17REZGEh8fz6BBgxgwYADbtm2jVq1aHDhwgNzcXPr374+7uztxcXHPTOe1bNkyOnToUGrarzeR/1QQyosYgAqCIJST2Ks5RQafBbKUWmKv5vBeJwsDtUqoys6dO0evXr106Yc8PDzIycnh4sWLjB07VlcvLy+v2HFKpfK5dfQpmh6pTXMn4h88oI6FJdcSEvAI/hoAlQRs335bd0y/fv1e/QYrqF27dpGSksLBgwcxMTHB1dWVXLH/tZhNmzaxYsUKPv/8c1xcXADw8fFh06ZNODg48P3335Odnc3jx49p3rw5np6eNG/enD59+gAFaa7u3bsHFHyJERwczG+//YZUKuXOnTu66+hL2WVlZUVsbCyffPIJUBAVXaVSkZmZSd/k2pCRRk5uLraOTTAyUuA9xB+AX3/9lU6dOlG3bl2g4He36LX69u2LVCrF0dGRxo0bExsby4kTJ/j999912y6ePHlCXFzcM9N5vf3223To0AEoPe2XGIAKlZkYgAqCIJSTnGyN/vIc/eWC8CZotVqsrKyQy+Wl1tFoNKXWMTY2RqPR6OqpVCrdc0WXI+ZbWpIrAS1aWjVuxJnQZahMTUrsAS2aHqmqePLkCba2tpiYmPDLL7/oBkpFWVpaolQqDdA6w5s2bRp3795lypQpJCUlMXHiRO7evYupqSkmJiZ88cUXODo6Ympqiru7O7m5uSgUCi5fvoypqSmxsbFMmjSJxMREZDIZHTt2xM7OjnHjxrFx40ZycnIIDAykX79+elN2abVanJyciIiI0D23YcMGHjx4QFBQULG2rm3W7IX3fRZdYl308bx58+jatWux544fP15qOq+i74nS0n4JQmUmdjULgiCUEzNz/X9yzczEn2LhzejYsSM///wz2dnZZGZmIpfLMTc3x8HBQffhW6vVcv369WLH1apVq9Q6b7/9NlevXgXg8OHDxQagRWmNjMisZ4dD61b8kfGEY/fuktrkHXKkUmJjY9/ULVcIAwcOJCYmhh49erBjxw6aNm1aok7Lli2RSqXIZDLWrVtngFYaTkhICPXr1+e7775Dq9Vy+fJlfvrpJxo3bkynTp2oXbs2Bw4cIDQ0lD179pQ4fsuWLfj7+1O/fn0OHDigC9Szf/9+/P0LZiuNjIyIjo7We/0mTZqQmprKhQsXgIIZ1IYNG7J//34ePXoEQFpaWokvDtq3b8/Zs2dJTU1FpVKViHq7f/9+NBoN8fHxJCQk4OTkhLu7O5s3b9a9T27fvk1WVhZdu3YlLCxM9yVEUlKS7tpFvWg9QahMxAyoIAhCOWnexoy0FHWxZbg1LSQ0b2NmwFYJVVmbNm3o27cvHh4e2Nra0q5dOwBWrVpFUFAQy5cvR61W4+PjQ6tWrYodW1odf39/PvvsM2QyGd26dXvmDKbG2Jispk34fuMGgmbNImPZUvLz8xk1ahTNmzd/o/duCIWRsm1sbIrNrumrY2JiUqYpbyora2tratSoQa9evXBycsLPz4/du3fTsmVLjI2NUavVJY7p0KEDS5cu5cmTJ9y7d4+RI0fi6+tLSkoKv/76KxKJhFOnTumWsT7N1NSUtWvXMmvWLDIyMnS/k4GBgQwZMgStVouxsXGxCOsA9evXJyAggH79+lG7du0S75kGDRrg5eXFkydPWLhwIWZmZgwdOhSFQkGvXr3QarXY2NiwYcMG3N3duXnzpm4Jes2aNVm5cmWJ2dbS6tna2r7yay4IhibRarXPr1X2tPfv33+pA2xtbcU3PlWA6MeqQ/Tlq9FFwc3RYGZWMaLgir6sGsqiH2fMmEGbNm0YPHhwGbVKKOpFo2BXh/ekq6srBw8eZOPGjbrIy1CQq1apVDJjxgw0Gg2Ojo7cvXsXhULBiBEjdKmu4uPjOXLkCBs2bCAkJIQbN27oXUJrSNWhH6sL0Zev5s/915Kny8UMqCAIQjmysDQSAYeECmnRokX8+uuvTJ069Y1eR/MwGfZuRfs4FUkdG/DxR2pn/0avWRGIKNgvJiMjg7feegupVEp4eDj5+fkl6iQkJNC4cWNGjhxJYmIiv//+O+7u7nz22WeMHj0aW1tb0tLSUCqVvF0k4BVUvlRY1fX9IlRtZTIA9fPz6wUsB4yAf2/fvn1hWZxXEARBEITyERgYSGBg4Bu9huZhMtqls+BhMgBagDuxaKb8q8p/qBZRsF/MiBEjGDNmDDt27Ch1iXdERAQ7d+7E2NiYevXqMXHiRKytrfUuoS06AK1sXwJU5/eLULW99hJcPz8/I+AG4AHcA84DQ7Zv3/7bMw4TS3CrKdGPVYfoy6pD9GXVUBn6UfPvULTnTpQol7i6Ix0VYIAWlZ/TR5+Q8rDkbF7dekZ07larWFll6MvK6NIZJYl3SwbNatjI5I18CfC6/Vid3y8VjXhPvpo3uQT3A+DW9u3b7wD4+fn9BPgAzxqACoIgCIJQzWgfp75UeVVSEAW75ABURMF+fcmZeWyNeURalgrrmib4t7XF3tK0RL3KlgqrOr9fhKqtLP7qNQQURR7f+7NMEAShXB06dIgbN26UWT1BEMqWpI7NS5VXJc3bmFHTovhEgIiC/fqSM/OYfURBdHwGV//IJjo+g9lHFCRn5pWoW9lSYVXn94tQtZVbECI/P78xwBiA7du3v3T4aGNjYxFyugoQ/Vh1VMS+PH78ODVr1qRz585lUq+6qIh9Kby8Z/VjREQE//vf//jqq69KPX7z5s1cvHiR5cuXv6kmov50Io/jb5H/IFFXZlS/IXU+nYhxFf8dtLUFm4HWXDqbSlaWmpo1jXmvow21rErO1In35ItbdSGW5Mziy2qTM1Xs+N8T/tmreKqfTu5W/Lz3Pk8y/krtUsvKmE7ub+nth9f1uv1Ynd8vFY14T5atshiAJgIORR6//WdZMdu3b18HFGZa1r7sOmqx9rpqEP1YdZRXX27evJktW7YA8OTJE95++21iYmJ0ufT2799PVFQU/v7+REREcPz4cebOncsPP/zAL7/8wtatW8nLy+Odd95hxYoVXLt2rUQ9gODgYFJSUjA3N2fx4sU0bdqUyZMnU6tWLWJiYnj48CHBwcF4e3u/8Xsub+J9WTU8qx87depEp06dntnPT548IScn583+Lhibovm/2UiKRPXU+Pjz2NgUqsnvYKv3jCn8+JWbl0GuntsW78kXl5yWqb/8cabe19DlQ/MSqbBK64fX9dr9KN4vFYZ4T76aP/eAllAWA9DzQDM/P793KBh4fgIMLYPzCoIgMHz4cIYPH45KpcLPz48xY8Ywfvz4EvVcXFzw8PBAJpPpBolWVlb4+/sDEBISwrZt2/jHP/5Rop6fnx8LFy7E0dGRS5cuERQUpEsQ/+DBA/bs2cOtW7f47LPPquQAVKjcFAoF/v7+dO7cmVOnTtGuXTv8/PwIDQ3l0aNHrFq1ihs3bnDlyhXmz59PSkoK06dPJzGx4LviOXPm4OLiUm7tldrZgwigIpQR65omQHbJcnMTvfUrWyos8X4RqqLXHoBu375d7efnNwH4mYI0LBu2b99+/bVbJgiCUMSsWbNwc3PD09PzhY+JjY1l0aJFZGRkoFQqcXd3L1FHqVRy8eJFxo4dqyvLy/tr71CvXr2QSqU4OTnx8OHD17sJQXhD4uPj2b59OwsWLKBPnz7s2bOHPXv2cPjwYVauXEnPnj11dWfNmsXo0aP54IMPSExMZOjQoZw4UTLSpiBUBv5tbbnxKLvYMlx7y4JARIIgVExlsgd0+/btB4ADZXEuQRCEp4WFhXHv3j3mz58PgETyVyCP3NzcUo+bMmUK69evp1WrVoSFhXHmzJkSdTQaDVZWVsjlcr3nMDX9a1/Q66atEoQ3xcHBgdatW/Po0SOcnJzo0qULEomEFi1aoFAoitU9efJksSBcmZmZKJXK8m6yIJQJe0tT5vRwKIiCm63C2rz0KLiCIFQM5RaESBAE4VVcuXKFtWvXsmvXLqTSgkiFdnZ23Lx5kyZNmnDo0CEsLAqWU1laWhb7IJ2ZmUn9+vVRqVTs3r0be3v7EvVq1aqFg4MDERER9O3bF61Wy2+//UarVq3K+U4F4dXVqFFD97NUKtV9cSKVSsnPL576Q6PREBERgZmZiL4qVA32lqYEuOnfayYIQsVTMeNOC4Ig/Gnjxo08fvyYQYMG4eHhwZdffklQUBAjRoygX79+1KtXT1fXx8eH77//Hk9PT+Lj4/nqq6/w9vamf//+NG3atNR6q1at4qeffkImk9GtWzcOHz5siFsVhHLh7u7Oxo0bdY+vXbtmwNYIgiAI1Y2YARWESurw4cPcuHGDCRMmGLopb9TSpUv1lusLBuTi4sLx48d1j//2t78xYsSI59YD2Lp1a4l6y5YtK/a4MPKuIFRmc+fO5euvv0Ymk6FWq3F1dSUkJMTQzRIEQRCqCYmB9jRp79+//1IHiPDHVYPox5ejVqsxNq6Y3xNVh75Mzswr2FeUpcK6ZtXdV1Qd+rI6EP1YdYi+rBpEP1Ydoi9fzZ9pWCRPl1fMT7aCUE0sXbqUXbt2UbduXRo0aICzszNRUVG8++67nD9/Hh8fHxwdHVmxYgV5eXlYW1uzatUq7OzsCAsL06VViIiIYOnSpUilUqysrNi1axexsbFMnTqVvLw8tFot69atw9HRkZ07d7Jhwwby8vJo374933zzDUZGRoZ+KSqc5Mw8Zh9RFImsmM2NR9nM6eFQJQehQvWSmfsHVx/uIFv1GHOTOrSx88WyRr3nHygIgiAIr0nsARUEA7l8+TIHDhxALpfz448/EhMTo3tOpVJx8OBBxo0bxwcffEBERASHDx/Gx8eH7777rsS5li1bxtatW4mKitLt7dqyZQsjR45ELpdz4MAB3nrrLW7evMm+ffvYs2cPcrkcIyMjdu3aVW73XJlsjXlULKw/QHKmiq0x4htQ4S8KhYLu3bu/9nliYmKYOXNmGbTo+TJz/+B4Qgh308/wMOt37qaf4XhCCJm5f5TL9QVBEITqTcyACoKBnD9/np49e+oiUXp4eOie69evn+7npKQkPv/8c/744w/y8vJo1KhRiXO9//77TJkyhb59+9K7d28AOnTowIoVK0hKSqJ37944Ojpy6tQprl69Sp8+fQDIycnB1lbkStMnLUulvzxbf7kgvI62bdvStm3bcrnW1Yc7UKqKDzaVqoIZ0U5vf1EubRAEQRCqLzEDKggVUM2aNXU/z5w5k88++4wjR44QEhKiN+9lSEgIgYGB3L9/n969e5OamsqAAQPYuHEjZmZmDBs2jFOnTqHVahk0aBByuRy5XM7JkycJCAgoz1urNKxrmugvN9dfLlRf+fn5fPXVV3Tr1o0hQ4aQnZ3NtWvX8Pb2RiaTMXLkSB4/fgyAr68v8+fPx8vLiy5dunDu3DkATp8+zfDhwwFISUnhk08+oVu3bnz55Zd88MEHpKamlll7s1WP9ZbnlFIuCIIgCGVJDEAFwUBcXFyQy+Xk5OSgVCqJiorSWy8jI0OXvzI8PFxvnfj4eN577z2++uor6taty/3790lISKBx48aMHDmSnj178vvvv9OlSxf279+v20iflpbGvXv33swNVnIFAYeKDzbtLQsCEQlCUXFxcYwYMYJjx45hZWXFgQMHmDx5MsHBwURFRdGiRQuWLFmiq69Wq4mMjGTOnDnFygstWbIENzc3jh07hpeXF4mJiWXaXnOTOnrLzUopFwRBEISyJJbgCoKBtGvXDk9PT2QyGXZ2drRs2ZJatWqVqBcQEMDYsWOpXbs2bm5uKBSKEnXmzZtHXFwcWq2WLl260KpVK1avXs3OnTsxNjamXr16TJw4EWtrawIDAxkyZAharRZjY2Pmz5/P22+/XR63XKnYW5oyp4dDQRTcbBXW5lU3Cq7wehwcHGjdujUAzs7OJCQkkJ6eTqdOnQAYNGgQY8eO1dUvXALv7Oys9wug//73v6xfvx6Abt26UadO2Q4M29j5kpJ1u9gyXAuTerSx8y3T6wiCIAiCPmIAKggGNG7cOAICAsjOzmbgwIE4Ozvj7+9frE7Pnj3p2bNniWMHDx7M4MGDAfj3v/9d4vkJEybozRHq4+ODj49PGd1B1WZvaUqAWwNDN0Oo4GrUqKH72cjI+1SzIwAAIABJREFUiPT09GfWNzU11dVVq9VvtG36WNaoR9fG07j6cAc5qseYiSi4giAIQjkSA1BBMKDAwEBu3LhBbm4ugwYNok2bNm/0eka5udRKfoBRnpp8U2Oe2Ncnv8iHZ0EQXp+VlRW1a9fm3LlzuLq6snPnTjp27PjCx7u4uBAREcH48eM5ceKEbv9oWbKsUU8EHBIEQRAMQgxABcGAVq9eXW7XMsrNxeZ2HCZ5f0ZxVYKJMovUJu+IQagglLFly5Yxffp0cnJyaNSokd69nqWZOnUqX3zxBTt37qRDhw7Uq1cPCwuLN9haQRAEQSg/Eq1Wa4jrau/fv/9SB9ja2uoCpwiVl+hHw6mTcJeaaSWXBmZZ1+Zx45KpXZ5H9GXVIfqyYsnNzcXIyAhjY2MuXLhAUFAQcrn8ucdVl36cPHkyMpkMb29vg50/LCwMd3d3XYC4slZd+rKqE/1YdYi+fDUNGjQAkDxdLmZABaGaMMrTv9estHJBEMqfNC+FxzH/j0+n/QeNVoKRuQ2LFy82dLOEp4SHh9OiRYs3NgAVBEGoysQAVBCqiXxTY1CWUi4IgsFJ81Koc38DtjapXPzBEwC1sQ2PGzigMXDbDCk8PJy1a9cC0LJlS4yMjDh37hzr1q3j4cOHBAcH4+3tjVarZd68eRw7dgyJRMKkSZN0AddWr17Nrl27kEgkdO/ena+//ppr167plkk3btyY0NDQEhGHly5dqkuX9f777xMSEkJkZCQxMTFMmDABMzMz9u3bh7m5ebm/LoIgCJWVyAMqCNXEE/v6qEyL57VUmZrwxL6+gVokCEJRFqlyjNWpxcqM1alYpD5/+W1VFRsby/Lly9m+fTtRUVH861//AuDBgwfs2bOHTZs28c033wBw4MABrl+/jlwu56effmLevHk8ePCAo0eP8vPPP7N//36ioqL4/PPPAZ6Zq7XQp59+yoEDBzh69CjZ2dnI5XK8vb1p27Ytq1atQi6Xi8GnIAjCSxIDUEGoJvJr1CC1yTtkWdcm18KCLOvaIgCRIFQgRuonL1VeHfzyyy94e3tjY2MDgLW1NQC9evVCKpXi5OTEw4cPgYL8qf3798fIyAg7Ozs6duxITEwMJ0+eZPDgwbqBorW1NRkZGSVytZ47d67E9U+fPo23tzc9evTg9OnT3LhxozxuWxAEoUoTa+8E4Snr169n8+bNtGnThlWrVpV4/tq1azx48IAePXoABR9QTExMcHFxKe+mvrT8GjVeKeCQIAhvXr5xrZcqr84Kc6kCvKlgijk5OXz99dccOHCAhg0bEhoaSm5u7hu5liAIQnUiZkAF4SmbNm1i27ZtegefANevX+fo0aO6x2fOnOHixYvl1TxBEKoopY0HamObYmVqYxuUNh4GapHhubm5sX//flJTC5Ymp6WllVrX1dWVffv2kZ+fT0pKCufOnaNdu3Z89NFHhIWFkZ2drTtH0VytgN5crYWDTRsbG5RKJZGRkbrnLCwsyMzMLNN7FQRBqC7EDKggFDFt2jTu3r3LsGHD6NevHwkJCcTGxqJSqQgICKBbt258++235OTk6JZ7bdmyBSMjI3bu3Mm8efNwdXU19G0IglAJaUzr8rjBP7BIlWOkfkK+cS2UNh5oTOsaumkG07x5cyZNmoSvry9SqZTWrVuXWrd3795cvHgRDw8PJBIJwcHB1KtXj3r16nH9+nV69+6NiYkJ3bt3Jygo6Lm5WmvXrs3QoUPp0aMHdnZ2tG3bVvecn58f06dPF0GIBEEQXoHIAyqUq8rQj66urhw8eJC1a9fi5OTExx9/THp6Ol5eXhw+fJiIiAiuXLnC/PnzAQgNDcXCwoJx48YZuOXlqzL0pfBiRF9WDaIfqw7Rl1WD6MeqQ/TlqxF5QAXhJUVHRyOXy1mzZg1QsBwrMTHRwK0SBEGo3p5Oy9K3b19WrPj/7N13XNVl+8Dxz2FvFRePOSoTNRVcJI4ClSWYEyEn5q7M3Iap/dwPKpqjcqTmqsDF0lQcOTIc9QhYPaiIilsRQQ7rcDi/P3g8eQScbK/369Wrc+7zHff3fMFzLu77vq5lZGdnU6VKFVasWEH16tVRKpVMmzaNmJgYFAoF48aNw8vLi88//5zo6GgyMzPx8vJi4sSJABw9epTZs2ejVquxt7dn/vz5GP8vSZtedpKMTAshRBGRAFSIQmg0GlavXs1bb72l0/7HH3+UUo+EEOLV9rAsS1hYGNbW1iQnJ6NQKAgPD0ehUPDDDz/wzTff8OWXX/LVV19haWnJgQMHCAoK4tChQ3h5eTFlyhSqVKmCWq3G19eXv/76izfffJNx48YRFBRE/fr1GTNmDA0bNuTSpUva+qyPlsgxzEzkfq0h2iD0RWbCNGjQgPPnzxf5eySEEGWdJCESohBOTk6sX79em2Hx7NmzAFhYWOgkn5BkFEIIUTIKKsty48YN7VrNb7/9Vlsq5ejRowwePFi778PRzPDwcNzd3XF3dycuLo7z588THx9P3bp1qV+/PpBXliU3NxeQ+qwFUavVpd0FIUQ5JgGoEIUYO3YsKpUKFxcXOnbsyIIFCwBo164d58+fx9XVldDQUFxdXdmzZw+urq4F1pETQghRPNLT0/H29ubixYtoNBq6detGUlIS3bp14/z584wYMUL7B8Jbt27Rq1cvpk+fzjvvvMP+/fu19T1HjBihs7b/Ucl3b9Lho93s/u0ql26k4Tp2Hy0+DOP9kSsKXJaxZcsWPD09cXFxYfjw4drsu1euXOH999+nc+fOBAQE6Ozz7bffavdZtGhRsbxPAwcOxMXFhU6dOhEaGkqbNm202YWjo6Px9vYGQKlUMm7cODp37oyLi4s2+2+DBg2YOXMmLi4u/P7772zfvh0vLy9cXV2ZPHmyBKVCiGcmAWg5tm/fPm2pkMDAQO1axYIcP36cQYMGlVTXitTNmzcZPnx4iZ3vxIkTWFtbY2pqyoIFCzhw4ACHDh1i48aNQN5f3IN3fs+s9T2xtPuT28Z7Cdn1A5GRkZIBVwghitHjZVkiIiJQKBSsWbOGgwcPcvnyZc6dO8esWbMYOnQoDg4OmJiYABAbG8vnn3/OW2+9xZ49e4iJieHAgQNERESwdetWqlSpwm+//caePXvYvn07enp63Llzh17jg/i/Ic3xbFubsUtPMtCjPv9Z3w0fTwemT5+er49dunRh9+7d7N+/n7feeosff/wRgBkzZjBo0CAOHDhAzZo1tdsfPnyYhIQEdu3axb59+4iJieHo0aNF+r4dOnQIGxsb9u/fz8GDB+nYsWOh2z46dXn//v20b98eyAtiW7Rowf79+6lSpQphYWGEhIQQGRmJvr4+O3bsKNI+CyEqLglAyzE3NzdGjx5dIufKyckpkfMUxMbGhjVr1pTa+R+XlnWbXy4HcCXlN+6k/82VlN/45XIAaVm3S7trQghRoT1alsXFxYW9e/eip6eHr68v7777LgYGBhgaGtK8eXM+++wz0tPTcXNzY+HChbzxxhu88847NGvWjOTkZCZPnszrr79O/fr1qVWrFosXL+bWrVtMnDgRPb28r0e+vr5MnfZ/OLfNK/8S9dcd+rq8QY6BNe8PmMLJkyfz9TEuLo6ePXvSuXNndu7cSVxcHACnTp2iR48eAPTu3Vu7/eHDhzl8+DBubm64u7sTHx/PhQsXivR9a9SoEUeOHGHu3LmcOHECKyurQrd9fOpy5cqVAdDX18fLywuAY8eOERsbi6enJ66urhw7dowrV64UaZ+FEBWXJCEqoxITE+nfvz8tW7bk9OnTNG/eHB8fHwIDA7l79y4rVqzg3LlzBU4ZSkhI4PPPPycpKQl9fX1ttsD09HSGDx9OXFwcdnZ2LF++HIVCQUxMDDNnzkSpVGJtbc2SJUuoWbMm3t7evP3225w6dYru3buzfft27TkuXrzI5s2bsbe3Z9q0aTq1Mt3d3QkKCtLp26BBgxg1ahTdunXjxx9/5Ouvv6ZSpUq8/fbbGBkZMXfuXMaOHYulpSXR0dHcuXOHL774gq5du5KYmIifnx8HDx4kLi6O8ePHk52drU0SZGNjw8iRI7lx4wa5ubl89tlndO/evdDrelmxd7ahVOkGm0rVbWLvbKNt7Y9f+vhCCCEK5+Pjg4+Pj/Z5cnIyBw8e5IcffqBu3bo0atQIyFufv3TpUgDtZxLkjfDdu3ePUaNG8eDBA3bt2kVa1m0M3oily6CG3LqsZNbczwkLC6NZs2Yc+vUP2jmO+N+az61kmNuRXdODXEXBQdy4ceNYu3YtTZo0ISgoiN9++037mkKRrxoBGo2G0aNHM3DgQG1bUZd8qF+/Pnv27OHgwYMsWLCADh06YGBgoF3nmpWV9dRjGBsbo6+vr+1znz598Pf3L7I+CiFeHTICWoZdunSJkSNHcuTIES5cuEBISAghISHMmDGD5cuXF7rfp59+yuDBg9m/fz+hoaHaoOvs2bPMnDmTX375hcuXL3Pq1ClUKhXTpk1j9erV7NmzB19fX521KSqVip9//plRo0YRGRlJZGQkkyZNws7OjtatW7N06VLat2/Prl272Lp1K7NnzyY9Pb3Qvl2/fp2vvvqK8PBwQkJC8v2V99atW4SEhLBhwwbmz5+fb/9NmzYxdOhQIiMj2b17N//6178KnFr0tOt6GRmq+wW2ZxbSLoQQonjcvHkTU1NTevfuzahRo/jPf/7D7du3OXPmDABpaWlPnMHTvHlzfvvtVyJiZnHp3nGO7D1DjYYZ/HI5AIUCFi9eTHx8PMvX/MQDmw9o+U4HNp8wJteoKjt27Chw2UVaWho1a9ZEpVKxc+dObbuDgwOhoaEAOtNVnZ2dCQoKQqlUAnDjxg1u3y7aGTWPv0+xsbHUrl1bG5Q/XOcJ8N577/H9999rn9+/n/+zrUOHDkRERGiD5OTkZK5evVqkfRavlsTERDp16vTM23t7exMdHV2MPRLFSUZAy7A6derQuHFjAGxtbenQoQMKhYJGjRqRmJhY4D5paWncuHGDLl26AGjXvkDeB+3/CsLSpEkTEhMTsbKyIi4ujg8++ACA3NxcatSood2nW7duOse/ePEic+bMYevWrRgaGj53rcxTp07h6OhIlSpVAOjatSsXL17Uvu7h4YGenh62trbcuXMn3/6tWrVi2bJl2mt88803adSoEbNmzWLu3Lm4uLjQpk0b/vvf/z7xul6GqWHlAttNCmkXQghRPP773/8yZ84cFAoFhoaGzJ8/H41Gw7Rp08jMzMTExISgoKBC969ZsyY+oxxZOHYfaKBxm6o0bVcdpeo2uZoc9PX1+frrr/nwww8xNzdnzpw5jBs3jpUrV2pn1jxu0qRJdO3alapVq9KiRQttEqRZs2bxySef8M033+Dm5qbd3snJifPnz2s/b83MzNi8eTOVKlUq1vcpMzOTCRMmsHDhQtq2bavd9rPPPmPq1Kl06tQJPT09xo8fj6enp87xbG1tmTx5Mn379kWj0WBgYMDcuXOpXbt2kfVZCFFxSQBahj1MGQ+gp6eHkZGR9vGLZJt7uD/kreXIyclBo9Fga2tLeHh4gfuYmZlpHyuVSkaNGsXChQu1o6qF1cqMiYnRTu2BZ5ve83gfH5Y/eVTPnj1p0aIFBw4cYODAgQQEBNChQ4d8U4s8PDyeeF0vo1l1b5LS43Wm4Zob1qBZde8iP5cQQojCOTs74+zsnK89IiJC57mvry++vr7a5w+TygG06vQadR3zj2R+FzkCyPss/uGHH7TtW7duzbfthAkTtI/9/Pzw8/PLt03dunV1PpOmTJmifTxs2DCGDRumfV7UU3ALe5+OHTuWr+3RqcuPelizNCUlhaioKHJycpg4cSKOjo5FGiyLV1dOTg6jR48mNjYWW1tbli1bxsqVK4mMjCQzM5PWrVsTEBCgncq+fft2Jk6ciFqtJjAwEHt7e959913CwsKoWrUqubm5Os9F2SFTcCsYCwsL/vWvf7Fnzx4gL/B7mAK+IPXr1+fevXucPn0ayJty+zBhwuPGjx+Pr6+vzpSjwmpl1qlThz///JPc3FyuXbumnQ7VunVroqKiuH//Pjk5Oezevfu5ru/y5cvUq1ePoUOH4u7uzt9//13g1KLnua7nZWFcA+d6U6hbqS01zBpTt1JbnOtNwcK4aEZYhRBClJzSntWSkpLC3r172b59O3v37iUlJaVEzvsiUlJStImVrl27RlxcHDt37izTfRblR3x8PH5+fhw+fBhLS0s2bNjA4MGD2b17NwcPHiQjI4PIyH/q7z58Pm/ePCZMmICenh69e/fWTnE/evQob7/9tgSfZZCMgFZAy5YtY8qUKSxatAgDAwNtEqKCGBkZsWrVKmbMmEFqaipqtZphw4bRsGFDne2uXr3Krl27uHjxIj/99BMAixYtYuzYsXz55Ze4uLiQm5tLnTp12LhxIw4ODtStWxdnZ2caNGhAs2bNAHjttdf49NNP8fLyokqVKtSvXx9LS8tnvrbw8HC2b9+OgYEBNWrU4NNPPyU6Ojrf1KJnva4XZWFcQxIOiTIpKSkJPz8/srOzmT17tpQGEuIpSnNWy8OALjU1Vdt248YNhgwZUuznfhFRUVE6fQVITU0lKioKd3f3UuqVqChq1aqFg4MDAL169WLdunXUqVOHb7/9loyMDO7fv0/Dhg21U9i7d+8OgKOjIw8ePCAlJQVfX1+GDBnC8OHD+emnn3QSlomyQ1HQNMcSoLl+/fpz7VDU01FE6ahWrRqXL1/G3NycnJwchg4dygcffKBdsyrKD/mdLJtCQ0M5evRogcXs1Wq1Novlox69lxqNBo1Goy1DIcoP+Z18cWlZeZnMM1X3MTGsTLPq3iUyq2Xv3r0Fzs6xs7MrcMpsadu+fXuBeR5q165Nr169SqFHZVtx/04+WiXgWXh7ezN9+nTs7e2LrU8vKjExkd69e2tLGx07doz169dz+vRpdu/ezWuvvUZgYCCQN+Xd29ubcePGaevUOjg4cODAAaysrBgwYACjRo1i0qRJHDt2rMDPvecl/76+mP/lnsmX/ltGQEWJCwwM5OjRo2RlZeHk5ISHh0eRn0M/KwvLm7fQz85BbWTAA5uaqB9ZUytEebJkyRJ27NhB1apVqVWrFnZ2dnh4ePDFF1+QlJSEqakpCxcuJDMzkzlz5pCZmUl0dDRhYWHY2dkxYMAAjh49yrx58/jPf/6jTcrSt29fhg8fzqVLl+jSpQstWrQgNjaWTZs2ER4eTnh4ONnZ2Xh4eDBx4kQSExMZMGAA77zzDqdPn8bGxgZ7e3sqV67MqFGjSvldEuLFldaslocJih734MGDEu7Js7GwsCiw3dzcvIR7Iiqia9eucfr0aVq3bk1ISAgODg6cPn0aa2trlEolu3bt0taiBQgLC6N9+/acPHkSKysrbX3bvn37MmbMGHr37l0kwacoevInblHiZsyYQWRkJEeOHGH27NkF1kV7GfpZWVjHJ2CWnIKxUolZcgrW8QnoP2MiJCHKkjNnzrB7924iIyPZvHmzNu385MmTmT17Nnv27GH69On4+/vTtGlTJk6cSLdu3YiMjMTU1JT09HRatGjB/v37MTExITg4mIiICMLDw/nhhx+067YTEhLw8/Pj0KFDxMfHk5CQwK5du9i3bx8xMTFERUXl287KykqbmEQI8fwKC+ieZ2lKSXJ0dNR+yX/IysoKR0fHUuqReJi4x8nJieHDh5ORkUFMTAy9e/fGw8ODfv36cevWLe3227dvx9XVlU6dOvGf//wHyKsTP378eLy8vHBzc2Pv3r2lci3169dnw4YNODk5kZKSgp+fH/369aNz587069cv38itsbExbm5ufP755zqzftzc3FAqlTqJx0TZIiOgosKxvHkLw2yVTpthtgrLm7e4X69uKfVKiBdz6tQp3N3dtSWVXF1dyczM5Pfff2fkyJHa7bKzswvcX19fX/sX45MnT+Lh4aHNbt2lSxdOnDjB66+/Tu3atWnVqhWQV/vwzJkz7NixA0NDQ/T19Tl16hQLFy5EX1+f6dOns3DhQuzs7Pjll1+057p06VK+UdnHM2QLIf7h6OjIjRs3dNZVWllZ0alTJ51M8mVFpUqV6NmzJ1FRUSiVSszNzSULbimLj48nMDAQBwcHxo8fz/fff8/PP//M+vXrqVq1KqGhoQQEBLB48WLgn8Q9UVFRTJgwgYMHD2prui9evJiUlBS8vLx49913dSohFLc6depw5MiRfO1TpkzRyRj90LZt2wo91l9//cXbb78tnz9lmASgosLRzy646Hhh7UKUNxqNBisrK51sgIUxNjZ+pilID79onDlzhoSEBL788kt8fHxwd3dn4MCBHDx4kPHjxzN9+nTtiGvnzp11viRPnjyZf//737z55pv88ccf+Pv7F1iyQgiRp7CAztrausyuN6tUqZIkHCpDHk/cs3z58ifWQS8ocU9hNd0bNGhQwlfz4h6WB9q5cyfHjx/nq6++Ku0uiSeQAFRUOGojA1AW0i5EOePg4MCUKVMYPXo0arWa/fv3M2DAAOrUqUN4eDjvv/8+Go2Gv/76iyZNmjzxWG3atGHcuHGMHj0ajUbDnj17WLZsmc42p06dwsnJiR07duDr64urqyt37tzh9OnTfPHFF1y5coUpU6bkG3FVKpXPPCorhPiHBHTiZTy+jMnCwuKJddAf316hUBRa0728eDSb9Ntvv83bb7/NlStXSElJkdH5MkrWgIoK54FNTVRGhjptKiNDHtjULKUeCfHimjdvjpubGy4uLgwYMIDGjRtjaWnJihUr+Omnn3BxcaFjx47s27fvqcdq1qwZffr0wcvLi65du9K3b1+aNm2ab7t69erRo0cPunXrxrZt2wgNDcXCwoJNmzbx+uuvExkZyeHDh3X2yc3N1Y7KPvzv8W2EEEIUrYeJewBCQkJo2bLlE+ugh4WFAegk7imspnt58aTyQKJskiEhUeGojY25V/8NyYIrKoxRo0YxYcIEMjIy6NWrF3Z2dtStW5ctW7bk29bX11cn8cLjSYJGjhypM0oJ8Prrr2vT+D8ccQ0NDaVv3764u7szYMAAdu/ezZkzZzh48KB2xHXUqFEolXnTDSwtLV9oVFYIIcSLe5i4Z8KECdja2jJkyBCcnJwKrYP+MHFPTk6OtqxJYTXdy4vCskk//HwSZY8EoKJCUhsbS8IhUWFMnjyZc+fOkZWVRZ8+fWjWrFmxnevREdfq1avrjLhOmjKB+YtmoMpR8a5rM2a9tVRn3xUrVuDv78/SpUvJycmhe/fuEoAKIUQxKSxxT9OmTdmxY0e+9sIS95iamrJgwYIi719JkfJA5Y8EoEIIUcZ9/fXXJXq+gkZcrWuaMHBmXZQqE+12v1wOYOToKVgY10A/Kws7Dez5v9ky60AIIcoJvewkzO9Fop/zALWBJUprV3KNqpZ2t55LYdmkpTxQ2SUBqBBCCB0Fjbj+dvUblKrbOtspVbeJvbONDtWHYh2f8E/5IyUYKtO5V/8NCUKFEAAkJibi5+enne4vSp9edhKVr6/DIOeets0wM5H7tYaUqyBUygOVPxKACiGE0FHQiGuG6n6B22aq7kvtXSFEscrJycHAQL6yFjXze5E6wSeAQc49zO9F8sDmg1Lq1YuRbNLli2TBFUII8VSmhpULbDcxrCy1d4V4BWzcuBFXV1dcXV1xdHTE29tbp05kREQEY8eOBeDOnTsMHToUFxcXXFxcOHXqlM6xLl++jJubG2fOnOHSpUv0798fDw8PevbsyYULF4C8xDhTpkyha9euzJkzp+Qu9BWin/PgudqFKCry5yQhhBBP1ay6N0np8TrTcM0Na9Csujfqm5lSe1eICm7QoEEMGjQIlUqFj48PI0aM4JNPPilw2+nTp+Po6MjatWtRq9UolUpSUlIAuHDhAh9//DFLliyhSZMm+Pj48O9//5s333yTP/74A39/f7Zu3QrAjRs3CA0NRV9fv8Su81WiNrB8rnYhiop8OxBCCPFUFsY1cK43hdg728hU3cfEsDLNqntjYVyDBzZZGCrTdabhSu1dISqmGTNm0L59e9zc3Ard5tdff2Xp0rws2fr6+lhZWZGSkkJSUhJDhgzhu+++w9bWFqVSye+//65TGio7O1v7uGvXrhJ8FiOltSuGmYk603BzDKxRWruWYq/Eq0ACUCGEEM/EwrgGbWt/nK9dau+KkvY8CW2CgoKIiYlh7ty5JdCzii0oKIirV69q30uFQqF9LSsr66n7W1pa8tprr3Hy5ElsbW3Jzc3FysqKyMjIArc3MzMrmo6LAuUaVeV+rSHlPguuKH9kDagQQoiX9rD2blKDN7lfr64En0JUMDExMaxatYrly5ejp5f39bF69eqcP3+e3Nxc9uzZo922Q4cObNy4EQC1Wq0tj2FkZMTatWvZtm0bO3fuxNLSkjp16hAeHg6ARqPhzz//LOEre7XlGlXlgc0H3K89nAc2H0jwKUqEBKBCCCGEKLceJrT5448/eP/993Fzc6Nbt27aZDaP2r9/P++//z737t1jy5YteHp64uLiwvDhw8nIyADyRlf79OmDi4sLPj4+XLt2raQvqUxav3499+/fp0+fPri6ujJx4kT8/f3x8/OjW7du1KhRQ7vtrFmzOH78OJ07d8bDw4Nz585pXzMzM2PDhg2sWbOGffv2sWLFCn766SdcXFzo2LEj+/btK43LE0KUIIVGoymN82quX7/+XDtUq1aNu3fvFlN3REmR+1hxyL2sOOReVgyv0n18OAV39erV2oQ2devWxdTUFAMDA44cOcKmTZtYs2aNdgpuhw4dWL16NevXr6dy5crcu3cPa2trAAICAqhevTpDhgzBz88PLy8vfHx8+Omnn9i3bx/r1q0r0et7le5lRSb3seKQe/liatWqBaB4vF3WgAohhBCi3Hk8oc21a9cYO3YsCQkJKBQKVKp/kmL9+uszMWj7AAAgAElEQVSvREdH8+OPP2JpmZfhMy4ujgULFpCamopSqcTJyQmA33//ne+++w6A3r17SwmQEqZMUxMXm0lmRi4mpno0bGaCuYUkIhKiIpEpuEIIIYQodx5NaAOwcOFC2rVrx8GDB/n+++91kuLUq1cPpVLJxYsXtW3jxo1jzpw5HDhwgHHjxj1TEh1RvJRpaqJ+SePaFRVJd9Rcu6Ii6pc0lGnq0u6aEKIISQAqhBBCiHLn8YQ2Dx48wMbGBoDg4GCdbWvXrs2aNWv47LPPiIuLAyAtLY2aNWuiUqnYuXOndtvWrVsTGhoKwI4dO2jTpk0JXZGIi80kXam7NCxdqSEuNrOUeiSEKA4SgAohhBCiRGVlZeHr64urq6s22HsRjya0adKkCfPnz8fNzY2cnJx827711lusWLGCkSNHcunSJSZNmkTXrl3p0aMHb731lna7OXPmEBQUhIuLC9u3b2fWrFkv3D/xfDIzcgtuzyy4XQhRPkkSIlGi5D5WHHIvKw65lxVDebqPv//+OwsWLCAoKKi0u6Kll51UZuohlqd7WZT++E3JtSuqfO2v1TWkZVvzUujRy3lV72NFJPfyxUgSIiGEEEIUq61bt7Jq1SoAGjduzPvvv8+yZcvIzs6mSpUqrFixAoVCwZgxY0hKSsLV1ZU1a9aQmprKzJkzUSqVWFtbs2TJEmrWrFli/dbLTqLy9XUY5NzTthlmJnK/1hCpi1iCGjYzITkpR2carpm5gobNTEqxV0KIoiYBqBBCCCFeWlxcHEuXLiUsLAxra2uSk5NRKBSEh4ejUCj44Ycf+Oabb/jyyy9ZuHAhK1euZOPGjahUKsaMGcP69eupWrUqoaGhBAQEsHjx4hLru/m9SJ3gE8Ag5x7m9yJ5YPNBifXjVWduoY+js0VeFtzMXExMJAuuEBWRBKBCiAonOjqabdu2MXv27NLuihCvjF9//ZWuXbtqa2tWqVKFv//+m48++ojbt2+TnZ1N3bp18+0XHx9PXFwcH3yQF+jl5uZSo0aNEu27fs6D52oXxcfcQr9cTrd9Fm3atOHnn3/G2tqabt26ERYWVmznSklJYefOnQwePLjYziHEi5IAVAhR5mk0GjQaDXp6z5Y3zd7eHnt7+2LulRDiaaZPn86IESNwc3Pj+PHjBY5qajQabG1tCQ8PL4Ue5lEbWD5XuxAvqziDT4DU1FQ2btwoAagokyQLrhCiTEpMTOTdd99lzJgxdOrUiQkTJtClSxc6duzIokWLtNudOXOGbt264eLigpeXF2lpaRw/fpxBgwYBEBgYyPjx4/H29qZt27asXbtWu+/27dvx8vLC1dWVyZMno1ZLrTkhXlT79u2JiIjg3r28qazJycmkpqZqS6Ns3bq1wP3q16/PvXv3OH36NAAqlUpbKqWkKK1dyTGw1mnLMbBGae1aov0QFceQIUNwdHSkY8eObN68Od/rDRo0AOD48eP07t2bDz/8kLZt2zJv3jx27NiBl5cXnTt35tKlSwAkJSUxfPhwPD098fT05NSpU0Dhn3Hz5s3j8uXLuLq6Mnv2bG7dukWvXr1wdXWlU6dOnDhxomTeCCEKICOgQogyKyEhga+++opWrVqRnJxMlSpVUKvV+Pr68tdff/HWW2/x0Ucf8e2339K8eXMePHiAiUn+ZBUXLlxg69atKJVK3n33XQYNGsSlS5cICwsjJCQEQ0ND/P392bFjB3369CmFKxWi/GvYsCFjxozB29sbPT09mjZtyoQJExg5ciSVKlWiffv2JCYm5tvPyMiIVatWMWPGDFJTU1Gr1QwbNoyGDRuWWN9zjapyv9aQMpMFV5R/gYGBNGjQgMTERLy8vPD09Cx027/++otffvmFypUr065dO/r27cuuXbv47rvvWLduHbNmzWLGjBkMHz6cd955h2vXrtGvXz8OHz4MFPwZN3XqVOLi4oiMjARg5cqVODk58dlnn6FWq8nIyCiR90GIgkgAKoQos2rXrk2rVq0ACA8PZ8uWLajVam7dusX58+dRKBTUqFGD5s2bA2BpWfB0uc6dO2NsbIyxsTHVqlXjzp07HDt2jNjYWO2XgszMTKpVq1YyFyZEBeXj44OPj49Om7u7e77t2rVrR7t27bTPmzZtyo4dO4q9f0+Sa1RVEg6JIrNu3ToiIyPJycnh+vXrJCQkFLqtvb29NutzvXr1cHJyAqBRo0YcP34cgKNHj3Lu3DntPmlpaSiVSqDgz7jHNW/enAkTJpCTk4O7uztNmzYtsmsV4nlJACqEKLPMzMwAuHLlCqtWrWLXrl1UrlyZsWPHkpmZ+czHMTY21j7W19dHrVaj0Wjo06cP/v7+Rd5vIcSzK0v1N4UoCsePH+fo0aMcOXKE9PR0vL29ycrKKnR7IyMj7WM9PT3tcz09PXJycoC85Fzh4eEFzvIp6DPucY6Ojmzfvp0DBw4wbtw4RowYITN+RKmRNaBCiDLvwYMHmJqaYmVlxZ07dzh06BCQt3bs9u3bnDlzBsj7i/DDD+un6dChAxEREdrC0snJyVy9erV4LkAIUaCH9TdN06IxyryIaVo0la+vQy87qbS7JsQLe/DgAZUqVcLMzIwLFy7wxx9/vPQxnZycWL9+vfb52bNnn7i9ubk5aWlp2udXr16levXq9O/fn379+hEbG/vSfRLiRckIqBCizGvSpAlNmzalXbt2JCUlaacnGRkZ8e233zJt2jQyMzMxMTEhKCjomY5pa2vL5MmT6du3LxqNBgMDA+bOnUvt2rWL81JEBTVw4EBWrFhBpUqVSrsr5YrU3xQVkbOzM5s2bcLOzo7XX3+dli1bvvQxZ8+ezdSpU3FxcSEnJ4c2bdoQEBBQ6PbW1tY4ODjQqVMnOnbsSMOGDVm5ciUGBgaYm5uzdOnSl+6TEC9KodFoSuO8muvXrz/XDtWqVdOOVIjyS+5jxVEU9zInJwcDg2f/O1hiYiJ+fn4cPHjwpc4rdMnvZcVQHu9j5atrMMq8mK892+RN7tceXgo9KhvK470U+cl9rDjkXr6YWrVqASgeb5cRUCHECwkMDCQnJ4ehQ4fy5Zdf8tdff7F161aOHTvGTz/9xN69ezl//jwAERER7N+/n6+++oqxY8dibGzMn3/+SevWrYmPj+fmzZtAXoA5a9Ysevfuzbx58/jtt9/Izs7Gz8+PgQMHFlnfb6ZlsyX6LsnpKqqYGdLfvho2FkZP31EI8sr3rFu3juzsbFq0aMH8+fNp164dP//8MyYmJowcOZIbN26Qm5vLZ599Rvfu3Uu7y2WW1N8UovjJZ54oayQAFUK8kA4dOhAQEMDQoUOJiYkhOzsblUrFyZMnadOmDXv37i103xs3bhAaGoq+vr62LSYmhvHjx+Ph4cGPP/6IpaUlu3fvJisrix49euDk5IRCke+PaM/tZlo2Xx5I5Gaa6n8tGZy7m8HMznXkA1k81fnz5wss3/PQoUOHsLGxYdOmTUBeMXhROKW1K4aZiTrTcKX+phBFRz7zRFkkSYiEEC+kZcuWxMbG8uDBA4yMjGjVqhXR0dGcOHGCNm3aPHHfrl276gSf9+7dY8yYMaxYsQIrKysOHz7Mtm3bcHV1pWvXriQnJz8xhf3z2BJ995EP4jw301RsiZapNeLpHi3f4+rqyrFjx7hy5Yr29UaNGnHkyBHmzp3LiRMnsLKyKsXeln0P629mWNiTbfImGRb23K81RLLgClFE5DNPlEUyAiqEeCGGhobUqVOH4OBgWrduTePGjTl+/DiXLl2iQYMGOqOVj6eff1heBUCtVvPRRx8xbtw4GjVqpG2fM2cOzs7OOvsVVMT+eSWnqwpuzyi4XYhHFVa+Jzg4GMjLzLxnzx4OHjzIggUL6NChA+PGjSuNrpYbUn9TiOIjn3miLJIRUCHEC2vTpg0rV66kTZs2tGnThk2bNtG0aVMUCgXVq1fn/Pnz5ObmsmfPnkKPMW/ePBo3bqyzTs7JyYmNGzeiUuV9QMbHx5Oenl4kfa5iZlhwu2nB7UI86mnle27evImpqSm9e/dm1KhRxMbGMnbsWCIiIkqry0KIV5h85omySEZAhRAv7J133mHZsmW0bt0aMzMzjI2NeeeddwDw9/fHz88Pa2tr7O3tUSqVBR5j5cqVNGzYEFfXvDVfkyZNol+/fiQmJuLh4YFGo8Ha2pp169YVSZ/721fj3N0MnSlJNhZ5SRmEeJrCyvc89N///pc5c+agUCgwNDRk/vz5OrX7hBCiJMlnniiLpAyLKFFyHyuO8nwvtRkBM1RUMZWMgOX5XpY1S5YsYceOHVStWpVatWphZ2fHf//7X1xcXOjatStLliwhMjKSzMxMWrduTUBAAAqFgoSEBD7//HOSkpLQ19dn1apVvP7663z77beEh4eTnZ2Nh4cHEydOLPTcch8rDrmXFUNZuY/ymffyysq9LG+kDIsQolzKvXMTQreguX8PRWVr6N4fveo2L3VMGwsjJrSvVUQ9FCLPgeOn2LA1FI8v1lLJWEHojMHY2dnpbDN48GDtmtBPP/2UyMhI3Nzc+PTTT/nkk0/o0qULmZmZaDQaDh8+TEJCArt27UKj0TB48GCioqJwdHQsjcsTQpRT8pknyhoJQIUQZVbunZtolsyAO3l1QjUAF+PIHTfrpYNQIYrSzbRs/h20H4MGjvx9PxcAg/rvkJqVo7Pd8ePH+fbbb8nIyOD+/fs0bNiQdu3acePGDbp06QKAiYkJAIcPH+bw4cO4ubkBkJ6eTkJCggSgQgghyjVJQiSEKLtCt2iDT63/jYgKUZZsib7Lg6xcnbZ0lZr/XP9n7XNmZiZTp05l1apVHDhwgH79+uXLEP0ojUbD6NGjiYyMJDIykl9//ZW+ffsW2zUIIUpHgwYNnvh6YmIinTp1KqHeFG7fvn2sWLECgD179nDu3Dnta0FBQdy8ebOwXbUkKZsACUCFEGWY5v6952oXorQkp6uweL0JKX/9Rq4qG3VWBvf/jiJd9U9Q+jDYtLa2RqlUsmvXLgAsLCz417/+pc0WnZWVRUZGBs7OzgQFBWkTeN24cUPWIAkhSo2bmxujR48G8gegW7du5datW6XVNVHOyBRcIUSZpahsTUFp0hSVrUu8L0I8SRUzQ8zrNKLS2+34c8kwDC2qYGrzJpUrWUFy3peySpUq0a9fPzp37kz16tWxt7fX7r9s2TKmTJnCokWLMDAwYNWqVTg5OXH+/Hm6desG5NXPXb58OdWqlUz2ysTERPz8/Dh48OAzbf/xxx9z7tw5fHx86NSpEx999BEKhYLVq1fz+uuvF29nhaggCks8lpOTw+jRo4mNjcXW1pZly5ZhampKTEwMM2fORKlUYm1tzZIlS6hZsybe3t60aNGC48ePk5KSQmBgIG3atCEuLo7x48eTnZ2NRqNh9erVGBoa0r9/f1q2bMnp06dp3rw5Pj4+BAYGcvfuXVasWEGLFi0ICgoiJiaGHj16EBkZSVRUFEuXLqVHjx5ER0czevRoTExMCAsL4/z58wX2SwiQAFQIUZZ17w8X43Sn4Va3yWsXogx5WOpA7eTDa25+qLMzubh6PIM92tKxzSfa7aZMmcKUKVPy7f/mm2+ydetWnTZlmpqWTfry9he+mJjq0bCZCeYW+sV+LS/i9u3bREdH8+uvvwKwYsUKvLy8GDt2bCn3TIjyo7DEY6+99hrx8fEEBgbi4ODA+PHj2bBhA0OHDmXatGmsX7+eqlWrEhoaSkBAAIsXLwbygtZdu3Zx4MABFi9eTFBQEJs2bWLo0KH06tWL7Oxs1Go1d+/e5dKlS6xatYrFixfj6elJSEgIISEh7Nu3j+XLl+uUQnNwcMDV1VWb3Rvg0KFDTJ8+HXt7e1Qq1RP7JYQEoEKIMkuvug2542YVeRZcIYqajYURMzvXod+QUdxJvIgmJ5sBH/jQsU3LFzqeMk1N1C9ppCsfzgFQk5yUg6OzRakEoZcvX2b48OH8+9//5ssvvyQrKwsTExMWL17MW2+9Rb9+/bh58yaurq506dKFjRs3oq+vz7Fjx9i2bRvbt29n3bp1ZGdn06JFC+bPn4++vj6HDx9m0aJFZGdnU69ePZYsWYK5uXmJX58QZUFhicdee+01atWqhYODAwC9evVi3bp1ODs7ExcXxwcffABAbm4uNWrU0B7P09MTADs7O65evQpAq1atWLZsmTbx2ZtvvglAnTp1aNy4MZBX77hDhw4oFAoaNWpEYmLic11HfHz8E/slhASgQogyTa+6DQybUNrdEOKpbCyMOBi87ukbPoO42MxHgs886UoNcbGZtGxbsgHahQsX+Pjjj1myZAl169Zl586dGBgYcOTIEQICAlizZg3r16/Hz8+PyMhIIC+Bkrm5OaNGjeL8+fOEhYUREhKCoaEh/v7+7Nixg86dO7N06VKCgoIwMzPj66+/ZvXq1doyNUK8ah4mHhs4cKBOe2JiIgqFbilFhUKBRqPB1taW8PDwAo9nZJRX61NfX5+cnLyM3D179qRFixYcOHCAgQMHEhAQQL169TA2Ntbup6enp91XT08PtVr93NfxpH4JIQGoEEIIUcZkZuQW3J5ZcHtxSUpKYsiQIXz33XfY2tpy7do1xo4dS0JCAgqFApVK9dRjHDt2jNjYWO1oTGZmJtWqVeP333/n3LlzdO/eHQCVSkWrVq2K9XqEKMucnZ1ZuHAhvXr1wtzcnBs3bmBoaAjAtWvXOH36NK1btyYkJAQHBwfq16/PvXv3tO0qlYqLFy/SsGHDQs9x+fJl6tWrx9ChQ7l27Rp///039erVe+6+WlhYaBOkAZibm5OWlgbwQv0SrxYJQIUQQogyxsRUD8g/6mBiUrLJ6y0tLXnttdc4efIktra2LFy4kHbt2rF27VoSExPx9vZ+6jE0Gg19+vTB399fp33fvn289957fPPNN8XVfSHKlcISj+nr61O/fn02bNjAhAkTsLW1xc/PDyMjI1atWsWMGTNITU1FrVYzbNiwJwZ64eHhbN++HQMDA2rUqMGnn36qDRyfR/fu3Zk0aRJr165l9erV+Pj48Pnnn2uTED1vv8SrRaHRFJRjsthprl+//lw7VKtWTdLPVwByHysOuZcVh9zLsif/GlAwM1c8cQ1oUd/Hh1lwIyIi6Nevn/Zxr1698PLyIjAwkODgYE6cOJEvY25gYKB2Cu65c+f48MMPCQ0NpVq1aiQnJ6NUKjE1NcXDw4Pg4GDeeOMN0tPTuXHjBvXr1y+yayiv5HeyYnjV72Pu/+p2V4QcDq/6vXxRtWrVAlA83i4joEIIIUQZY26hj6OzBXGxmWRm5mJiUnpZcM3MzNiwYQN9+/bFxcWF+fPns3TpUjp37vxM+9va2jJ58mT69u2LRqPBwMCAuXPn0qpVK5YsWcInn3xCdnY2AJMnT5YAVIgKIPfOTTRLZmiz2GsALsaRO25WuQ1CRdGREVBRouQ+VhxyLysOuZcVg9zHikPuZcXwKt/H3O8C0Zw4nK9d0cYJvXKYWPBVvpcvQ0ZAhRBCCFEmpKSkEBUVRVpaGhYWFjg6OlKpUqXS7pYQooho7t97rnbxainZbAZCCCGEeKWlpKSwc+dO4uLiuHbtGnFxcezcuZOUlJTS7lq5MHDgwKe+V97e3kRHR+drP3v2LAcOHCiurgmhpahs/Vzt4tUiAagQQgghSkxUVBSpqak6bampqURFRZVSj8qXTZs2vfBo8Z9//qlNFCVEsereHx5f61ndJq9dvPIkABVCCCFEiSms5MOjNQVFnu3bt+Pl5YWrqyuTJ09GrVbTpk0b7t3Lm8a4ZMkS3n33XXr06MHHH3/MypUrtftGRETg5eVFhw4dOHHiBNnZ2SxatIiwsDBcXV0JDQ0trcsSrwC96jYoxs1C0cYJGjZD0cYJhSQgEv8ja0CFEEIIUWIsLCwKbDc3Ny/hnpRt58+fJywsjJCQEAwNDfH392fHjh3a18+cOcPu3buJjIwkJycHd3d37OzstK/n5OSwa9cuDhw4wOLFiwkKCmLixInExMQwd+7c0rgk8YrRq24D5TDhkCh+MgIqhBBCiBLj6OiIlZWVTpuVlRWOjo6l1KOy6dixY8TGxuLp6YmrqyvHjh3jypUr2tdPnTqFu7s7JiYmWFhY4OrqqrO/p6cnAHZ2dly9evWF+iDrTYUQxUECUCGEKMMSExPp1KlTsR7/p59+KrbjC/G4SpUq0bNnTxo2bEjt2rVp2LAhPXv2lCy4j9FoNPTp04fIyEgiIyM5evQoEyY8+2iSkZERAPr6+uTk5LxQH2S9qRCiOEgAKoQQrzAJQEVpqFSpEu7u7vTq1Qt3d3cJPgvQoUMHIiIitLUHk5OTdUYyHRwciIyMJDMzE6VSyf79+596TAsLi0LX4Mp6UyFESZEAVAghyricnBxGjx6Nk5MTw4cPJyMjg5iYGHr37o2Hhwf9+vXj1q1bAGzZsgVPT09cXFy02wKMHTuWiIgI7TEbNGgAwLx58/j1119xdXVl9erVqNVqZs+erT3Gpk2bSv6ChRDY2toyefJk+vbti4uLC3379tX+ngM0b94cNzc3XFxcGDBgAI0bN8bS0vKJx2zXrh3nz5/PFxQ+ut40MjISfX39Qtebbt68Od+U24frTWfOnMnixYsxMjJi4sSJdOvWjcjISLp3715E74oQoiKQJERCCFHGxcfHExgYiIODA+PHj+f777/n559/Zv369VStWpXQ0FACAgJYvHgxXbp0oX//vDT3AQEB/PjjjwwZMqTQY0+dOpV169bx3XffAbB582YsLS3ZvXs3WVlZ9OjRAycnJ+rWrVsi1/qqelgbc/DgwRw/fpyVK1eycePG0u6WKGXdu3fPF7ydOHFC+3jUqFFMmDCBjIwMevXqpU1CtG3bNu021tbW2n2qVKnC7t27853n0fWmAJmZmVSrVk37+qPrTYFiWW8qhHh1SAAqhBBlXK1atXBwcACgV69eLF++nLi4OD744AMAcnNzqVGjBgBxcXEsWLCA1NRUlEolTk5Oz3Wuw4cP8/fff7Nr1y4AHjx4QEJCggSgxSw1NZWNGzcyePDg0u6KKEcmT57MuXPnyMrKok+fPjRr1izfNso0NXGxmWRm5GJiqkfDZiaYW+jrbPNwvam/v79Oe3Bw8DP1oyjWmwohXh0SgAohRClKTEzEz8/vick6FAqFznMLCwtsbW0JDw/Xtu3bt48VK1awefNm1q5dS5MmTQgKCuK3334DwMDAgNzcXCAvYFWpVIWeb86cOTg7O7/EVYnnNW/ePC5fvoyrqyuGhoaYmZkxfPhw4uLisLOzY/ny5SgUCmJiYpg5cyZKpRJra2uWLFlCzZo1S7v7opR8/fXXT3xdmaYm6pc00pWa/7WoSU7KwdHZQicI7dChAx9++CHDhw+nWrVqJCcn69RldXBwYMqUKYwePRq1Ws3+/fsZMGDAE8/9pPWmQohXm6wBFUKIMu7atWucPn0agJCQEFq2bMm9e/e0bSqVinr16jF69GjS0tKoWbMmKpWKnTt3ao9Ru3ZtYmNjgbxg9WEA+viXRCcnJzZu3Kh9PT4+nvT09BK5zlfZ1KlTqVevHpGRkUybNo2zZ88yc+ZMfvnlFy5fvsypU6dQqVRMmzaN1atXs2fPHnx9fQkICCjtrosyLC4285HgM0+6UkNcbKZOW0muNxVCCBkBFUKIUqZWq5k0aRKnT5/GxsaG//u//2PMmDHs3bsXgLp16zJo0CCqV69O5cqViYqKwsDAgBEjRqBUKqlVqxb29vaYm5szadIkunTpglKpxMDAgLNnz3Lq1Cn69++Ph4cHW7ZswdzcHIVCwdq1axk0aBD6+vraKb6VKlVCT08Pd3d3IG/92Lp160rtvXlVNW/enFq1agHQpEkTEhMTsbKyKnTqtRAFyczILbg9M397Sa03FUIICUCFEKKUJSQk8PXXX7Nw4UJGjhzJ2bNnsbS05OzZszRt2pRu3bpRs2ZN+vXrR4cOHdixYwd169bl448/Ji0tjY0bNxIUFERMTAx+fn6cPHkSPz8/3nnnHa5du0a/fv04fPiw9v9bt25FqVTy7rvvMmjQIL766ismTpzId999h6GhIf7+/rRs2ZI+ffqU9lvzynq4pg7+WVen0WjyTb0W4klMTPUAdf52k+efAPcs601vpmWzJfouyekqqpgZ0t++GjYWRgUcTQjxKpMAVAghSlmdOnVo2rQpkJdFMjExkX79+hEcHEzjxo0JDw8nIiKCCxcuUK9ePW1CoB49erB58+Z8xzt69Cjnzp3TPk9LS9Ou5+rcuTPGxsYYGxtTrVo17ty5o82A6ebRheSMHLKysjiXbsy7XbrLl8cSYm5u/tT1cvXr19dOvW7dujUqlYqLFy/SsGHDEuqlKG8aNjMhOSlHZxqumbmChs1MnvtYT1tvejMtmy8PJHIz7eH68gzO3c1gZuc68u+IEEKHBKBCCFHKjI2NtY/19fXJzMzE09OTxYsX0759e5o1a4a1tTXXr19/puPl5uYSHh6uLZnwpHOp1Wo0Gg1duvfibvO+2i+POcCXBxLly2MJsba2xsHBgU6dOmFiYqJTAuMhIyMjVq1axYwZM0hNTUWtVjNs2DAJQEWhzC30cXS2yMuCm5mLiUnBWXCLwpbou48En3lupqnYEn2XCe1rFfn5hBDllwSgQghRBpmYmODs7Iy/vz+LFi0C8kbALl++TGJiInXq1CEsLKzAfZ2cnFi/fj0fffQRgHYqb2E6derErAVLqGfTGUOLKuSkp6LOyuAmNeXLYwkqbIRp7ty52sdNmzZlx44dJdUlUQGYW+jTsq15sZ8nOb3gzNrJGYVn3BZCvJpeKgD18fFZCLwPZAPxwIfBwcH3i6JjQgjxquvZsyc///yztpanqakp8+bNo3///piZmWFvb1/gfrNnz2bq1Km4uLiQk5NDmzZtnpgttXHjxtj3HM7JNVNAk4tC301/pnMAACAASURBVIC6PcZgXKWmfHksQ9KybhN7ZxsZqvuYGlamWXVvLIwlCZEoG6qYGQIZ+dtNDUu+M0KIMu1lR0AjAf/g4OAcHx+fAMAfmPLy3RJCiFdDnTp1dGqAjho1Svv41KlT+Pr6oq//z3S59u3bc+TIETQaDVOnTtVmovT19cXX1xfIm865cuXKfOeaMGGCzvNHz9uqYxeUb7TPt498eSwb0rJu88vlAJSq29q2pPR4nOtNkSBUlAn97atx7m6GzjRcG4u8RERCCPGolwpAg4OD9z3yNArwfrnuCCGEABg6dCiXL18mODhYp33Lli1s3boVlUpF06ZNGThw4EudJy3rNn/8vZY3rVOJuenE/Uwz7Wvy5bHsiL2zTSf4BFCq8kZE29b+uJR6JcQ/bCyMmNm5Tl4W3AwVVUwlC64QomBFuQZ0CBBUhMcTQohX1tq1awtsHzFiBCNGjCiSczw+qvbeGxf485YHxvqNqWZuJl8ey5AMVcGrWzILaReiNNhYGMmacSHEUz01APXx8dkP2BTw0hfBwcGh/9vmC/KSJm55wnFGACMAgoODC8zw98SOGhg89z6i7JH7WHHIvSz//vh7rc6omoVxKm3qBtOghjNujWU1RVlS+U5N7qT/na+9kkVN7e+h/E5WHHIvKwa5jxWH3MuipdBoNE/f6gl8fHwGAyOBzsHBwenPuJvmWcsJPFStWjXu3r37nL0TZY3cx4qjot/LxMRE/Pz8dNZJvoh9+/Zx7tw5Ro8eXUQ9KzoHE+YVGNTUMGtMxzemlkKPRGEeHa3OSFPxx8FbuPVuqbMGtKL/Tr5K5F5WDHIfKw65ly+mVq1aAIrH2/Ve5qA+Pj4ewGSg23MEn0II8cpwc3Mrk8EngKlh5QLbTQppF6XHwrgGzvWmULdSW8zUr3Nyd7IkIBJCCFEuvVQACqwALIFIHx+fMz4+PvnTLgohRDmlVquZNGkSHTt2pG/fvsTFxeHu7q59/eLFi9rnBw4c4L333sPDw4Pp06czaNAgAIKCgvjiiy8ASEpKYvjw4Xh6euLp6cmpU6cACAwMZPz48Xh7e9O2bVud9Z/bt2/Hy8sLV1dXJk+ejFqtLrLra1bdG3ND3QDG3LAGzapLPrmyyMK4Bm1rf8yhTSncuf6Anl37M3v2bL799ls8PT1p1aqVtmYsFO/PjhBCCPGiXioADQ4Ofis4OLhOcHBw8//9N+rpewkhRPmQkJCAn58fhw4dwsrKirNnz2JpacnZs2eBvODS19eXzMxMpkyZwubNm9mzZw9JSUkFHm/GjBkMHz6c3bt3s2bNGiZOnKh97cKFC2zZsoVdu3axePFiVCoV58+fJywsjJCQECIjI9HX12fHjh1Fdn0PR9Ua1HCmhllj6lZqK6Nq5cDUqVOpV68ekZGRvPfeeyQkJLBr1y5OnTpFTEwMUVFRxf6z8ypKTEykU6dO+donTpzIuXPnSqFHQghRPhVlFlwhhKhQ6tSpQ9OmTQGws7MjMTGRfv36ERwcTOPGjQkPDyciIoILFy5Qr1496tatC0CPHj3YvHlzvuMdPXpU54tqWloaSqUSgM6dO2NsbIyxsTHVqlXjzp07HDt2jNjYWDw9PQHIzMws8iQIFsY1cGs8Rda2lFOHDx/m8OHDuLm5YWBgQGpqKgkJCfz999/F/rMj8jw66iyEEOLpJAAVQohCGBsbax/r6+uTmZmJp6cnixcvpn379jRr1gxra2ueNalabm4u4eHhmJiYPPVcarUajUZDnz598Pf3f/mLERWSRqNh9OjRDBw4UCdJxrp16+Rnpxjk5OQwevRoYmNjsbW1ZdmyZQwcOJDp06djb29PSEgIy5cvR6PR0LlzZ+30eyGEEP942TWgQgjxSjExMcHZ2Rl/f398fX0BqF+/PpcvXyYxMRGAsLCwAvd1cnJi/fr12ucPp/IWpkOHDkRERGiDiuTkZK5evVoUlyHKMXNzc9LS0gBwdnYmKChIO5J+48YN7t69Kz87xSQ+Ph4/Pz8OHz6MpaUlGzZs0L528+ZN5s6dS3BwMPv27ePMmTPs2bOnFHsrhBBlkwSgQgjxnHr27IlCocDJyQkAU1NT5s2bR//+/fHw8MDc3BwrK6t8+82ePZvo6GhcXFxwdnZm06ZNTzyPra0tkydPpm/fvri4uNC3b19u3bpVLNckyg9ra2scHBzo1KkTR44coUePHnTr1o2WLVsyYsQI0tLS5GenmNSqVQsHBwcAevXqxcmTJ7WvRUdH07ZtW6pWrYqBgQG9evUiKiqqtLoqhBBllkzBFUKIAtSpU0enBuioUf/kWDt16hS+vr7o6+tr29q3b8+RI0fQaDRMnToVOzs7AHx9fbUjpdbW1qxcmT9Z+IQJE3SeP3re7t27071796K5KFFhfP311zrPhw0blq9OnfzsFD2FQvHE50IIIZ5ORkCFEOI5DB06lG3btjFs2DCd9i1btuDq6krHjh158OABAwcOfKnz6GUnYXnzJypfXYPlzZ/Qyy44s64QBZGfn+Jx7dq1/2fv3uNyvvsHjr86H64UOXVbsS1it0N2oMhKOhFyKobb4TanTbudbY73Y4g5pBkzszlvP3JWTimMDWGMmm2ROyS1mdLh6lzX74/mmiiU6rqq9/Ov+nwP1+d7fa643t/v5/N+8+OPPwKwb98+9dNQgPbt2xMZGUlycjIFBQXs27ePTp06aaqrQgihteQJqBBClMGjNTofNXbsWMaOHVshr6Gbe5+6dzegn5+sbjPIjudBk1EUGtavkNcQNZd8fiqPra0tmzdvZurUqdjZ2TFixAgiIiIAaNy4MbNmzcLPz0+dhOjRusGi5ggODiYqKoqAgIDnPsbBwYHDhw9jaWlJixYtuH79eiX2UAjtJgGoEEJoGUVyeLHgAUA/PxlFcjjpVu9oqFeiupDPT+WwsbHh1KlTT7Tv2rVL/XPfvn3p27dvVXZLCCGqHZmCK4QQWkYvP71M7UI8Sj4/VS81NZWwsDB2795NWFgYqampmu6SeA67d++mZ8+eeHh4MGPGDAoKCmjRogULFizA1dWVQYMG8dNPP+Hr60unTp04evSo+ti7d+/i6+uLk5MTK1asULePGjWK7t274+rqytdff/3U11cqlQwcOBAvLy/c3NwICwtTbwsKCuLtt9+mb9++vP/+++r8ATdv3lQnvOvXrx+xsbEV/K4IUfnkCagQQmiZAv06ZWoX4lHy+alaqamp7N27l7S0NHVbYmIi/fr1w8LCQoM9E09z/fp1QkJC2LdvHwYGBsycOZM9e/aQmZmJk5MTc+fO5d1332Xp0qVs27aNa9euMWnSJDw9PQG4fPkyx44dw8TEhJ49e+Lm5oa9vT2BgYHUq1ePrKwsfHx8cHZ2xtLSssQ+GBkZsX79eurUqUNycjK9e/fG09OTK1eucOjQIcLDw8nPz8fLy0ud2G7GjBl88sknvPrqq1y6dImZM2eyc+fOKnvfhKgIEoAKIYSWUVp6YJAdX2waZb6+JUpLDw32SlQX8vmpWpGRkcWCT4C0tDQiIyNlDagW++GHH4iOjsbb2xuA7OxsGjRogKGhIa6urgC0atUKQ0NDDAwMeO2114rV0n377bfVgWWPHj04f/489vb2bNiwgcOHDwNFSavi4uJKDUBVKhWffPIJ586dQ0dHh6SkJO7du8eFCxfw8vLC2NgYAA+Por9dpVLJxYsXGTdunPocubm5FfzOCFH5JAAVQggtU2hYnwdNRqFIDkcvP50C/TooLT0kgYx4LvL5qVoZGRkltiuVyiruiSgLlUqFn58fM2fOLNa+du1adXkdXV1djIyM1D/n5+er9yupJM+ZM2f4/vvvCQ0NxcTEhMGDB5OTk1NqH/bs2cP9+/c5fPgwBgYGODg4PHX/wsJCzM3NCQ8PL/P1CqFNZA2oEEJooULD+qRbvcMD6zGkW70jwYMoE/n8VB0zM7MS2xUKRRX3RJRFly5dOHDggLp2bkpKSrEnnM/y/fffk5KSQlZWFmFhYXTo0IH09HQsLCwwMTEhNjaWc+fOPfUc6enpNGjQAAMDA06fPq1+/Q4dOhAeHk52djZKpVKdablOnTrY2NgQGhoKFAXRV69eLc/lC6FR8gRUCCGEEKKcHB0dSUxMLDYN19zcHEdHRw32SjyLnZ0dM2bMYPDgwahUKvT19ctUVqV9+/aMGTOGxMREBgwYgL29Pa1atWLr1q24uLhga2uLg4PDU8/Rv39/RowYgZubG+3ataN58+bqc3t6euLu7k7Dhg157bXXqFOnaA336tWrmTlzJitXriQ/P58+ffrQunXr8r8RQmiAjkql0sTrqu7evVumAxo0aKC+SyWqLxnHmkPGsuaQsawZZBw1JzU1lcjISJRKJQqFAkdHxxdKQCRjWTO8yDg+/CxlZWXRv39/li5din3LJjK1XkPkb7J8mjRpAqDzeLs8ARVCCCGEeAEWFhaScEhUqBkzZnDt2jVycnLw8/PDvmUT6t7dUCy5mEF2PA+ajJIgVFQ7EoAKIYQQQgihRT7//PNivyuSthcLPgH085NRJIeTbvVOVXZNiBcmSYiEEEIIIYTQYnr56WVqF0KbSQAqhBBCCCGEFivQr1OmdiG0mQSgQgghhBBCaDGlpQf5+pbF2vL1LVFaemioR0KUn6wBFUIIIYQQQosVGtbnQZNRkgVX1AgSgAohhBBCCKHlCg3rS8IhUSPIFFwhhBBCCCG0WHx8PN26ddN0N4SoEBKACiGEqFZ2795Nz5498fDwYMaMGRQUFPDRRx/Ro0cPXF1dWb58uXpfBwcHkpOLShdcuXIFX19fTXVbCFHLqFQqCgsLNd0NIbSOBKBCCCGqjevXrxMSEsK+ffsIDw9HT0+PPXv28OGHH3L48GEiIiKIjIzkl19+0XRXhRA1xKJFi9i0aZP698DAQNauXcsXX3yBt7c37u7u6htf8fHxvP322/znP//h9ddf5+7du7Ro0YL//ve/uLq6MnDgQO7fvw/At99+qz5+zJgxZGVlAXDv3j3effdd3N3dcXd358KFC8X6c+vWLTw9Pbl8+TKTJk3iwIED6m0tWrQA4MyZM/j6+jJmzBicnZ3x9/dHpVIBcOzYMZydnenevTtz585l+PDhlfbeCVESCUCFEEJUGz/88APR0dF4e3vj4eHBDz/8wO3btwkNDcXLywsvLy9iYmK4fv26prsqhKghfHx8CA0NVf8eGhqKpaUlcXFxHDx4kKNHjxIVFUVkZCQAcXFxjBgxgsuXL2NtbU1mZib29vacOHGCTp06sWLFCgB69OjBoUOHiIiIoHnz5mzbtg2AuXPn4ujoSEREBGFhYbRs2VL92rGxsYwZM4agoCDat2//1H7//PPPfPzxx3z33XfcunWLCxcukJ2dzYcffsg333zDkSNH1MGwEFVJkhAJIYSoNlQqFX5+fsycOVPddvv2bQYPHszBgwepW7cukyZNIjs7GwB9fX31FLicnByN9FkIUb21adOGP//8k6SkJO7fv4+FhQW//fYbJ0+exNPTE4DMzEzi4uJ46aWXsLa25s0331Qfr6uri4+PDwD9+/dn9OjRAMTExLB06VLS0tJQKpW4uLgAcPr0aVauXAmAnp4e5ubmpKamcv/+fUaNGsXXX3+NnZ3dM/vdvn17mjRpAkDr1q2Jj4/H1NSUZs2a0bRpUwD69u3LN998U0HvlBDPR56ACiGEqDa6dOnCgQMH+PPPPwFISUkhISEBExMTzM3NuXfvHidOnFDvb21tTVRUFAAHDx7USJ+FENVfr169OHjwICEhIfj4+KBSqfD39yc8PJzw8HBOnz7N4MGDATA1NX3quXR0dACYPHkyCxcu5NixY0yePPmZN8nq1KnDSy+9xPnz59Vtj95kKywsJC8vT73N0NBQ/bOenh75+fllu2ghKokEoEKUQjLOCaF97OzsmDFjBoMHD8bd3Z3BgwdjaGhImzZtcHZ2ZsKECXTo0EG9/5QpU5g3bx49evRAT09Pgz0XQlRnPj4+7N+/n4MHD9KrVy+6du1KcHAwSqUSgMTERPWNsccVFhaqb4Dt3buXjh07ApCRkUHjxo3Jy8tj79696v27dOnCli1bACgoKCAtLQ0oCijXr1/Prl271PtbW1sTHR0NwNGjR4sFoCWxtbXl1q1bxMfHAxASElKu90OUXXBwMLNnz9Z0N7SCTMEVQghRrfTp04c+ffoUa3t0utujHBwc+OGHH6qiW0KIGqxly5YolUqsrKxo3LgxjRs35vr16+qptaampqxatarEG12mpqb89NNPrFy5kvr167N27VoApk+fTq9evahfvz6vv/46GRkZAMyfP58ZM2awfft2dHV1Wbx4MY0bN1afa/PmzQwePBiFQsHQoUP597//jbu7O66urs98+mpiYsKiRYsYOnQopqam2NvbV+TbJMRz0XmYEauKqe7evVumAxo0aFDqnSVRfVSncYyPj+df//oXHTt25Mcff8TKyooNGzawZ88evv32W3Jzc3nllVf47LPPMDExYdKkSbi7u9OrVy+gKBPd9evX+f3333nvvfdIT0+noKCAxYsX4+DgoOGre3HVaSzF09XksUxNTSUyMpKMjAzMzMxwdHTEwsJC092qFDV5HGsbGcua4eE4Pvw+oC2USiUKhQKVSsWsWbN45ZVXGDt2rKa7pdWe9jeZmZnJuHHjSExMpLCwkIkTJ9KsWTPmzZtHZmYmRkZGBAcHc/DgQcLDw8nKyuLmzZv06NGDOXPmALBv3z5WrVqFSqXCzc1N/aT05MmTLF++nNzcXJo1a0ZQUBAKhQIHBwf8/PwIDw8nPz+fL7/8kubNm1fZ+/G8/lqDrPN4u0zBFeIpHmayO3HiBObm5hw6dKjUrHWl2bt3Ly4uLup1Iq1bt66i3gtRu6WmprJ3715iYmJISEggJiaGvXv3kpqaqumuCSGERqSmpjJv3jw6duxIhw4duH//PsOGDdN0t6q1EydOYGVlRUREBMePH8fV1ZX33nuP+fPnExERwfbt2zE2Ngbg6tWrfPHFFxw7doyQkBASEhJISkoiICCAHTt2cPToUS5fvsyRI0dITk5m5cqVBAcHExYWhr29PevWrVO/rqWlJWFhYQwbNkz9VL26kCm4QjyFjY0Nbdq0AaBdu3bEx8eXmrWuNO3bt2fq1Knk5+fj5eWlPp8QonJFRkaq1049lJaWRmRkJF5eXhrqlRCittGWp58Pb8o1a9aMZs2aAWBubk5ubi4mJiYa7l311apVK+bPn09AQADu7u6Ym5vTqFEjdZmcOnXqqPft0qUL5ubmQFFOg4SEBFJSUujUqRP169cHijIlR0ZGoqenx7Vr19RLTvLy8ootN+nRowdQ9P308OHDVXKtFUUCUCGewsjISP2znp4e2dnZTJ48mfXr19O6dWuCg4M5e/YsUHomOkdHR3bv3q3Ocjd27Fj8/Pyq/mKEqGUerqd63MOkIUIIUZvITbnKYWtry5EjRzh+/DhLly7Fycmp1H0fzUysq6v71MzEKpUKZ2dn1qxZU+L2h99R9fT0KCgoKGfvNUOm4ApRRqVlrSstE92dO3do2LAhQ4cOZciQIep9hBCVy8zMrMR2hUJRxT0RQgjNk5tylSMpKQkTExMGDBjA+PHj+emnn/jjjz+4fPkyUPS+Py3QbN++PZGRkSQnJ1NQUMC+ffvo1KkTb775JhcuXCAuLg4oWmt648aNKrmmyiZPQIUoo9Ky1pWWie7MmTOsXbsWfX19FAqFurj087py5Qq7du1iwYIFFX4tQtRkjo6OJCYmFrvjb25ujqOjowZ7JYQQmiE35SrHb7/9xsKFC9HR0cHAwIDFixejUqmYM2cO2dnZGBsbExwcXOrxjRs3ZtasWfj5+amTED18Ih0UFMSECRPIzc0FYMaMGdja2lbJdVUmyYIrqpSMY80hY1lz1OSxfJgF92HWR8mCK6oDGcuaQdvG8eEa0MdvyvXr16/G/rtYUbRtLKuL0rLgyhNQISpJRs4fRN/bRVbeA9LvwbKpB3nrzQ78+OOPtG/fnoEDBxIYGMiff/7J6tWrAZg3bx45OTkYGxuzYsUKmjdvrn6CumXLFs6ePcu8efMA0NHRYc+ePSgUChYuXMiJEyfQ0dHhP//5D3369OHMmTOsWLGCevXqERMTQ7t27Vi1ahU6Ok/8OyBEjWVhYSFrm4QQgqJ/D/v161drbsrVVIX3kmD/t6geJKNT1xL6DEW3oZWmu1UmEoAKUQkycv7gu1tLUOb9AUByeha3bt5i5eqlrFixAm9vb/bt28e+ffs4evQoq1atYuXKlezduxd9fX1OnTrFkiVL+Oqrr4qdd+3atSxatIgOHTqgVCoxMjLi0KFDXL16lfDwcJKTk/H29lZPMfz55585fvw4VlZW9OnThwsXLtCxY8cqfz+EEEIIoXlyU656K7yXhCpoHtxLAkAF8L8YCifPr1ZBqCQhEqISRN/bpQ4+H7K0Mia3/hV0dXWxs7OjS5cu6Ojo0KpVK+Lj40lLS2PcuHF069aNjz/+mJiYmCfO26FDBz7++GPWr19Pamoq+vr6nD9/nr59+6Knp0fDhg1xdHTkypUrQNHC9iZNmqCrq0vr1q2Jj4+vkusXQgghhBAVbP+36uBT7a8notWJBKBCVIKsvAdPtOkb6JL9V7uurq46Fbeuri4FBQUsW7aMzp07c/z4cTZt2kROTs4T5/D392fZsmVkZ2fTt29fYmNjn9qPR9N96+npPTULmxBCCCGE0F6qB8llatdWEoAKUQlMDOqW2G5cSjtAeno6VlZF0yd27NhR4j43b97ktddeY8KECdjb2xMbG4uDgwMhISEUFBRw//59zp07py5+LIQQQgghagadupZlatdWsgZUiErQtqEv9zNvFJuGq6ujT9uGvqUe89577zFp0iRWrlyJm5tbift8/fXXnDlzRj2N19XVFUNDQy5evIiHhwc6OjrMnj2bRo0aPfPpqBBCCCGEqEb6DIX/xRSfhtvQqqi9GpEyLKJK1aZxfJgFNzvvAcYGdWnb0Bczo0aa7laFqU1jWdPJWNYMMo41h4xlzSDjWHNo01hWpyy4UoZFiCpmZtSITtbvl+mYhzW6Ro4cWebX+/nnn/n9999xc3NDLyeHUzt3ERN7gyn/Hsm87dswtbBg/PjxZT6vEEIIIYTQDroNrWD0VE1344XIGlAhtEhaWhpbtmwp17FXr17l+PHj6OXkYHkjDr827ZjTtx+mKamYpDxARxIQCSGEEEIIDZMnoEJokUWLFnHr1i08PDxwdnamQYMGhIaGkpubS/fu3Zk2bRqHDx9m48aNBAcH88cffzBgwACCg4NZvnw52dnZXDx9mtn9fcnKyeHH2Ousfm8CegUFGKanA0WJjGbPns39+/cxMTFh2bJlNG/eXMNXLoQQQgghagMJQIXQIrNmzSImJobw8HBOnjzJwYMHOXjwICqVipEjRxIZGUmPHj04dOgQmzZt4sSJE0ybNo2XXnqJadOmERUVxdqR72KkVLIp/Gixc+vmFwIwY8YMPvnkE1599VUuXbrEzJkz2blzpyYuVwghhBBC1DISgAqhpU6ePMnJkyfx9PQEIDMzk7i4OBwdHVmwYAFubm688cYb9O3bt9hxBYb6oHzyfIX6uiiVSi5evMi4cePU7bm5uZV6HUIIIYQQQjwkAagQWkqlUuHv78+wYcOe2JaYmIiOjg737t2jsLAQXd2/l3OnWzXGQJlZbP8CPT1y6tShsLAQc3NzwsPDK73/QgghhBBCPE6SEAmhRRQKBRkZGQB07dqV4OBglMqix5mJiYn8+eef5OfnM3XqVNasWUOLFi1Yt24dAGZmZmRkZFBgZESy7SvkKEwp0Dcgs54FWfXqotLXp06dOtjY2BAaGgoUBblXr17VzMUKIYQQQohaRwJQIbSIpaUlHTp0oFu3bpw6dYq+ffvi4+ODm5sbY8eOJSMjg1WrVtGxY0c6duzIf//7X/7v//6P69ev07lzZ65fv46Hhwd7jhwhs74l2XXNedCsKSr9vyc7rF69mu3bt+Pu7o6rqytHjx59So+EEEIIIYSoODoqlUoTr6u6e/dumQ7QpgKwovxkHGsOGcuaQ8ayZpBxrDlkLGsGGceaQ8ayfJo0aQKg83i7PAEVohYpvJdE4deBFCyfTeHXgRTeS9J0l4QWc3BwIDk5ucLOFxgYyNq1awFYtmwZp06dqrBzCyGEEKJ6kCREQtQShfeSUAXNg7+CThXA/2IonDwf3YZWGu2bqH2mT5+u6S4IIYQQQgPkCagQtcX+b9XBp9q9pKJ2UeuNGjWK7t274+rqyjfffPPE9p07d+Lu7o67uzsffPABAPHx8fj5+eHu7s7AgQNJSEh4avujJk2axIEDBwA4ceIEzs7OeHl5MXfuXIYPHw4Uf2IK0K1bN+Lj4wHYvXs3PXv2xMPDgxkzZlBQUFCxb4gQQgghKoUEoELUEqoHJU+lLK1d1C6BgYEcOXKEQ4cOsWHDhmJTb2NiYli5ciU7duwgIiKC+fPnAzBnzhz8/PyIiIigf//+zJ0796ntJcnOzmb69Ols2rSJI0eO8Mcffzyzr9evXyckJIR9+/YRHh6Onp4ee/bsecF3QIiqtX79elxcXPD399d0V4QQokrJFFwhagmdupaUlHJMp65llfdFaJ8NGzZw+PBhAO7evUtcXJx62+nTp+nVqxeWlkWflXr16gFw8eJFvv76awAGDBjAwoULn9pektjYWJo2bcqrr76q3r+kJ7CP+uGHH4iOjsbb2xsoCmIbNGhQ5msWQpM2b97M9u3bHybpKBeVSoVKpSpWC1oIIbSdBKBC1BZ9hsL/YopPw21oVdQuarWTJ0/y/fffExoaiomJCb6+vuTk5Gi6W+jp6VFYWKj+/WGfVCoVfn5+zJw5U1NdE+KFfPjhh9y+fZthw4bh5+fH+fPnzETbDAAAIABJREFUuX37NsbGxixduhRnZ2cCAwNRKBSMHz8eKJqCvnnzZgCGDBnC66+/TnR0NFu3bsXa2lqTlyOEEGUit8yEqCV0G1qhM3k+Og4u0LItOg4u6EgCIgGkpqZiYWGBiYkJsbGxXLp0qdh2JycnDhw4oJ6Wm5KSAsBbb73F/v37AdizZw8ODg5PbS9J8+bNiY+P5+bNmwDs27dPvc3Gxobo6GgAoqOjuX37NgBdunThwIED6pT4KSkp3Llz54XeAyGq0pIlS2jcuDE7d+7kzp07tGnThoiICD766CMmTpz4zOPj4uIYMWIEJ06ckOBTCFHtyBNQIWoR3YZWMHqqprshtIyXlxdr1qzBxcUFW1tb3njjjWLbW7ZsyX/+8x98fX3R1dWlTZs2fPrppyxcuJDJkyezdu1aLC0tCQoKAii1vSQPn/gMHz4cExMTHBwcyMjIAMDb25tdu3bh6urK66+/rp6ma2dnx4wZMxg8eDAqlQp9fX0CAgLki7iols6fP89XX30FFN1cSUlJIS0t7anHWFtb8+abb1ZF94QQosJJACqEELWckZFRiesuz507p/554MCBDBw4sNh2a2trdu7c+cRxpbVPnfr3zY9PP/1U/bOrqyuurq4AnDlzRp351sTEhG3btpXY5z59+tCnT5+nXZYQ1VppU9ABTE1NNdElIYSoEBKACiGEqDb0cnKok/Q7ern5FBjqk27VmAIjI013S4hyc3BwYM+ePUyePJkffvgBS0tLNmzYwNatW1EqlRgYGPDyyy9z8+ZNFi1aRHR0NL///jtZWVmYmJhw8+ZNZs+ezf379zExMWHZsmU0b95c05clhBClkgBUCCGE1ujcuTOdO3cucZteTg6WN+IwyM0ralCCgTKTZNtXJAgVWmXLli1s3boVgPT0dKytrfnggw9Yvnw5ubm5NGvWDJWqKC/5kSNHMDIyYtWqVfzjH//A1dWVefPmYW1tjZ6eHosWLaJr165AUZboWbNm4eHhwaFDhxgwYAAzZszgk08+4dVXX+XSpUvMnDmzxBkIQgihLSQAFUIIUS3USfr97+DzLwa5edRJ+p0HzZpqqFdCPGn48OEMHz6cvLw8Bg4cyDvvvMPKlSsJDg7G1NSUzz//nNdeew1LS0t0dXUZMmQI77//PklJSbi6ujJ+/HimTJnC4MGDef311+nYsSMxMb+iaHWN63nn8Rn+Fjdu/oZSqeTixYuMGzdO/dq5ubkavHIhhHg2CUCFEEJUC3q5+WVqF0LT5s2bh5OTExYWFly7dk29bjkvL69YEiEfHx8Arly5wssvv4xCoUBfX5/+/fvz7bffkpOfTp5OGrdTzwKQlhtPasYfpGcNxtzcnPDw8Art97Jly3BwcMDZ2blCzyuEECBlWIQQQlQTBYYl3zMtrV0ITQoODubOnTtMmTIFlUqFs7Mz4eHhhIeH89133xEYGKje99GkQg0aNCAkJISsrCxycnK4efMmFs2TKVQVv9GSW6AkLvsINjY2hIaGAkU1cq9evfrCfZ8+fboEn0KISiMBqBBCiGoh3aoxeYYGxdryDA1It2qsoR4JUbKoqCi+/PJLVq1aha6uLm+++SYXLlwgLi4OgMzMTG7cuPHEce3btycmJob+/fvj7e1NQEAAXl5e6JuUPK02O+8Bq1evZvv27bi7u+Pq6srRo0efu5/x8fG4uLgwffp0XF1dGTx4MFlZWUyaNIkDBw4ARUmSli9fjpeXF25ubsTGxqqvYcqUKfTs2RNPT0/CwsLK+jYJIWopuW0shBCiWigwMiLZ9hXJgiu03saNG3nw4AF+fn4A2NvbExQUxIQJE9RrNGfMmIGtrW2x4xo3bsysWbNYs2YNOjo6jBw5ktmzZ3P2zhpmfOWg3s/Vr2jNs7FBXZpaN+Xbb78td1/j4uL4/PPPWbZsGePGjePQoUNP7GNpaUlYWBibNm1i7dq1LF++nJUrV+Lk5MSKFStITU2lZ8+evP3221IiRgjxTBKACiGEqDYKjIwk4ZDQekFBQSW2lxTcPay3m5qayt69exk5ciSjR4/mzz//VO/TtqEv9zNvoMz7Q92mMGiEraIfl84qyc4qxNhEl5ZtjVGY6ZWprzY2NrRp0waAdu3aER8f/8Q+PXr0UG8/fPgwAKdOnSI8PFxdtzcnJ4eEhARatGhRptcXQtQ+EoAKIYQQQmhYWloaGzZt5n4LTzLy72KmD0PtG2BlZoiZUSO6NvuQ6Hu7yM57gLFBXWwV/Yg+bUqm8mFm6AJS7ufj2NWsTEGo0SMzCPT09MjOzi51Hz09PQoKCoCi9abr1q2TmqNCiDKTNaBCCCGEEBo2b/5C4m7e5IuJ77B7xRxCDh7hv8fiGTri30yZMgUzo0bc+sGSc8G6dLJ+n9XLdvHRf32Yv7QPx05uASBTqSIm+skAsjK4uLiwceNGdT3Tn3/+uUpeVwhR/UkAKoQQQgihYTbeYzC0bELryeuwsHuL9LhokjLy+DXuDteuXQOKpus6OjoSFRXF0WO7+XDiNmZM/D9+iNzF7Tu/ApCdXVgl/Z00aRJ5eXnq5EdLly6tktcVQlR/MgVXCCGEEOIFrV+/ni1btnDv3j3ef/99/P39y3R8atbfZVbMXmnL7z/sIev3m9T5x8s0NM3n999/5+LFiyxYsIDt27fT2dETI6OihD+vt3Mn9n8XaWr9GsbGz/9swcbGhuPHj6t/Hz9+/BP7PFyjCkXJlHbt2gWAiYmJBJ1CiHKRJ6BCCCGqnfj4eLp16/ZE+7Rp09RPi0TN8zDBTVJSEmPGjAGKpn4eO3ZMk90CYPPmzWzbto1ffvmlzMEngIXJ388EDC0aUpCVQWrMBV5t8yYODg6EhoaiUCgwMzMDoEEjfUwVOsXOYarQoWVb4xe7kGdITU0lLCyM3bt3ExYWRmpqaqW+nhCi5pEAVIgySE1NZdOmTeU69vEvSUePHmX16tUABAYGqjMJLlu2jFOnTr1wX4VmPDqWz+uzzz6rpN7UPsuXL8fOzu6593+YUEVUL1ZWVnz11VcAXL16tdhTvOeRn5//7J3K4MMPP+T27dsMGzaMdevWMXv2bNLS0ujYsSOFhUVTYjMzM3nrrbfIy8vj5s2bDB06lO7du9OvXz9iY2MZ8pYNOnlZ6nMqmr7Gn6f38H7/bnTs2JG1a9fSsWNHoKg25/ETR7F30KN+ozyifz1O584OZU5AVFYPM/XGxMSQkJBATEwMe/fulSBUCFEmEoAKUQZpaWls2bKlXMc+/iXJ09OzxLvk06dPx9nZudx9FNXPqlWrNN2Faik/Px9/f39cXFwYM2YMWVlZ+Pr6cuXKFQBOnjxJ79698fLyYuzYsSiVSqDoy3tAQABeXl4cOHCg1P2E9nr4BDw3N5fly5cTEhKCh4cH+/fvJzMzkylTptCzZ088PT0JCwsDIDg4mJEjR+Ln58egQYMqtD9LliyhcePG7Ny5EwsLCwDMzc1p3bo1Z8+eBSA8PJyuXbtiYGDAjBkzWLBgAUeOHGHu3LnMnDmTVjaNce7sQNyqMeSe3MQbHTpipg8dWtvRtm1bHjx4gINDUS3Qtm3b4ufnx8BBPsxZMIh3Rw/Fb2iHSg0+ASIjI0lLSyvWlpaWRmRkZKW+rhCiZpE1oEKUwaJFi7h16xYeHh44OzvToEEDQkNDyc3NpXv37kybNo3Dhw+zceNGgoOD+eOPPxgwYADBwcEsX76c7Oxszp8/j7+/P9nZ2URFRREQEFDsNSZNmoS7uzu9evUiKiqKjz/+GKVSiaWlJUFBQTRu3BhfX19ef/11zpw5Q2pqKoGBgeovJqJyjBo1irt375KTk8O7777Lv/71L06cOMEnn3xCQUEBlpaW7NixA4Br167h6+tLQkICo0eP5t133y31HIsWLSI7OxsPDw9atmypfiounu3GjRsEBgbSoUMHpkyZwubNm9XbkpOTWblyJcHBwZiamvL555+zbt06Jk+eDEC9evUICwsjOTmZ0aNHl7qf0G6GhoZMmzat2L+lixcvxsnJiRUrVpCamkrPnj15++23AYiOjiYiIoJ69epVSf98fHwICQnBycmJkJAQRowYgVKp5OLFi4wbN069X25uLgAbvvwCgAYNGhTVAZ37HgAGBgbExsYCRU8hIyMjadCgAR9++CGOjo7qoLeyZWRklNguN22EEGUhAagQZTBr1ixiYmIIDw/n5MmTHDx4kIMHD6JSqRg5ciSRkZH06NGDQ4cOsWnTJk6cOMG0adN46aWXnviSFBwc/NTXysvLY86cOWzcuJH69euzf/9+lixZwooVK4Cipz8HDx7k2LFjrFix4pnnEy8mMDCQevXqkZWVRc+ePfHy8mL69Ons2bOHpk2bkpKSot43NjaWnTt3olQqefvttxk+fDgGBgZPnMPb25tZs2axceNGwsPDNXh11VOTJk3o0KEDAP3792fDhg3qbRcvXuTatWv06dMHKPp7evPNN9XbfXx8nms/Uf2cOnWK8PBw9VT4nJwcEhISAHB2dq6y4BOKZrp88sknpKSkEBUVhZOTE5mZmZibm5frb/7hFNhHn0ImJibSr1+/CglCp02bxtixY0udxv5w/enjFArFC7+2EKL2kABUiHI6efIkJ0+exNPTEyha3xMXF4ejoyMLFizAzc2NN954g759+5br/Ddu3CAmJoZ33nkHgMLCQho1aqTe7u3tDUC7du24c+fOC16NeJYNGzZw+PBhAO7evcs333yDo6MjTZs2BSj2pdbNzQ0jIyOMjIxo0KAB9+7do0mTJk+cIy4uDktLy6q/mBpCR0en1N9VKhXOzs6sWbOmxGNNTU2faz9R/ahUKtatW0fz5s2LtV+6dEk97lVFoVBgb2/PvHnzcHd3R09Pjzp16mBjY0NoaCi9e/dGpVLxyy+/0Lp162ee72lTYL28vF64v8uXL3/qdkdHRxITE4v1wdzcHEdHxxd+bSFE7SFrQIUoJ5VKhb+/P+Hh4YSHh3P69GkGDx4MFN2R1tHR4d69e+oEFOU5v52dnfr8x44dY9u2berthoaGAOjp6VV4Qg1R3JkzZ/j+++8JDQ0lIiKCNm3aPPXLopGRkfpnPT09CgoKSjxHTk5OVXS/xkpISODHH38EYN++feqnoQBvvvkmFy5cIC4uDii6QXTjxo0nzvG8+wntZWZmVmxqqIuLCxs3bkSlUgFFCeA0ycfHhz179qifugOsXr2a7du3q2toHj169LnOVZFTYDMzMxk2bBju7u5069aN/fv3q9dQ37lzBycnJ5KTkyksLKRfv36cPHkSCwsL+vXrR8uWLbG2tqZly5YV9vRVCFF7SAAqRBkoFAr1F4CuXbsSHBys/o8/MTGRP//8k/z8fKZOncqaNWto0aIF69atA578kvQstra2JCcnq79g5+XlERMTU8FXJJ5Heno6FhYWmJiYEBsby6VLl8jJySEyMpLbt28DFJuC+7zneMjAwIC8vLxKvYaayNbWls2bN+Pi4kJqaiojRoxQb6tfvz5BQUFMmDABd3d3fHx8Sgwsn3c/ob06d+7M9evX1UmIJk2aRF5enjq4q6palefOncPS0pJBgwYVW9vfq1cvEhIS6NSpk7qtadOmfPvtt0RERPDdd98995rjipwCe+LECaysrIiIiOD48eO4urqqt1lbWzNhwgQ++ugjvvzyS1q0aIGLiwsAFhYWeHl50b9/f7y8vCT4FEKUmUzBFaIMLC0t6dChA926dcPV1ZW+ffuq72qbmpqyatUqtm7dSseOHenYsSP//Oc/8fb2xs3Njc6dO/P555/j4eHxXDXiDA0N+fLLL5k3bx5paWkUFBQwevRoWrZsWdmXKR7TtWtXtm7diouLC7a2trzxxhvUr1+fpUuXMnr0aAoLC2nQoAHbt28v0zmOHj1KQkICQ4cOxd3dnbZt26qTEG3ZsgUTExP8/PwIDg7GxcUFKyurqrpkrWdjY1NiuaJdu3apf+7SpQuHDh16Yp9z584V+720/YT2uX79OlA0/g+ziterV++J8Ssp6Bw0aFCFZ78tr4x79zi09Rve692HEz9HEbh7F5u/+eaZx1XkFNhWrVoxf/58AgICcHd3fyKR3ZAhQzhw4ABbt2597ie0QgjxPHQeTlGpYqq7d++W6QB1RjhRrck41hwylpUnPz8fff2/7w/6+voyd+5c7O3tK+X1avNY6ubeR5Ecjl5+OgX6dVBaelBoWL9K+xAfH8+IESOeu5ZlaYliavM4lkQbxrYkejk5pJ0+S785s/h5zZd8F3WFZfv2sGHbNgr+mr5f0lgWFBSgp6enzoKrVCpRKBQvlAU3JSWF48eP83//93906dKF06dPq/+tycrKwtvbm5ycHPbu3Uvjxo1f+NprG/mbrDlkLMunSZMmADqPt8sTUCGqGb2cHOok/Y5ebj4FhvqkWzVWf2kRmpeZmcm4ceNITEyksLCQiRMnsmjRInr37s2JEycwNjZmTdBC2pjH8MkXIZgq6vDupEX0HzKOf/7zn1y4cIE+ffqov1xaW1tz5coV/P39MTY2JiQkhK5du3L48GEsLS25cuUKCxYsYNeuXWRmZjJnzhxiYmLIy8tj6tSpeHl5ERwcXCwD8/Dhwxk/fjxNmzZl0KBBnD59msLCQgYMGMCkSZPUU+1qOt3c+9S9uwH9/GR1m0F2PA+ajNKKQKU0z0oUI7R7bOsk/c7Er9ZxIzGR9v7vY6Cvh8LYmPGj3iU6/jbt2rVTr/d3cHDAx8eHU6dO8f7771O3bl2WL19Obm4uzZo1IygoCIVCUWrJrqdJSkqibt26DBgwAHNz82I5BgACAgLo168f1tbWTJ8+vdw1sIUQ4nGyBlSIakQvJwfLG3GYpqRipFRimpKK5Y049CSZjdYobV1VnTp1OHbsGP/+ly8LZn+AScYV9PJTMMhNpO7dDVCYR15eHocPH2b8+PHq8/Xq1Qt7e3tWr15NeHg4JiYmpb72ypUrcXJy4uDBg+zcuZMFCxaQmZlZ6v4P13l98MEHT6zzqg0UyeHFAhQA/fxkFMlVXxInPz8ff39/XFxcGDNmDFlZWXz//fd4enri5ubGlClT1EmrHiaKKSgoYNKkSXTr1g03NzdWrlwJwPr16+natSvu7u68915RHcmUlBRGjRqlrjH8yy+/AEXlhaZMmYKvry+dOnVi/fr1QMkJaqoTbRrbx+nl5vPJyH9j+49/cHn1GpaNGs1PN26w/H1/vvvuO27dusWZM2fU+z+sWfv222+ra9uGhYVhb2/PunXr1CW71q1bx5EjRxg0aBBLlix5Zj9+++03evXqhYeHB0FBQUycOFG97ezZs1y+fJkJEybQv39/DA0NpdSXEKLCyBNQIaqROkm/Y5BbPFmNQW4edZJ+50GzphrqlXhUaeuqHpbj+VcXQz6en1DsGP38ZPTyU4plySyPp9U/LM2QIUMICwurleu89PLTy9RemW7cuEFgYCAdOnRgypQpfPnll3zzzTcEBwdja2vLf/7zH7Zs2cKYMWPUx1y9epWkpCT11F19fX3y8/P5/PPPOXv2LEZGRqSmpgJFgWabNm3YsGEDP/zwAxMnTlTXoSypbu3DGylbt24FeKL0h7bTprF9XIHhk1+9Otq15B/WTXigq0vr1q25deuWer3/s2rWPqtkV2m6du1K165di7U9uob6wIED6p+//vrrsl2kEEI8hQSgQlQjerkll1sprV1UPVtbW44cOcLx48dZunQpXbp0Af6uUalXkP5E/UoAHVX+c9co1NfXV5f3ebSUS2n1D6OiooqVA3r0mKysLHWQqlQqS82yWRMV6NcpU3tlatKkibqMTP/+/fn0009p2rQptra2APj5+bF58+ZiAWjTpk25ffs2c+bMwc3NjQEDBpCcnMxrr72Gv78/3bt3p3v37gCcP3+er776CihKupSSkkJ6elEwVlLd2mclqNF22jS2j0u3akzejf8VazM0MiTdqmjK7OOltZ5Vs/bXX3/Fzs6O0NDQCutj4b0k2P8tqgfJ6NS1hD5D0W0oSdCEEBVDpuAKUY2UdOf8ae2i6iUlJWFiYsKAAQMYP3480dHRAISEhACw4/gNHP/Z8InjVDqlj+Gj5X+gaOpsVFQUAAcPHlS3l1b/0MbGhqtXr1JYWEhCQgKXL19WHxMQEMDgwYOZNm0a06dPL+9lV0tKSw/y9S2LteXrW6K09Kjyvjx+U+J5ksrUrVuX8PBwOnXqxNatW9VTt7ds2cLIkSOJjo7G29v7mXWCS6pb+/BGSqtWrVi6dClBQUHluCrN0aaxfVyBkRH5/2xFWnYOOQoF2XXMyDU1feZa/tJq1lZ0ya7Ce0moguahOncSYqJRnTuJKmheUVAqhBAVQAJQIaqRdKvG5BkaFGvLMzRQ3zkXmlfauqrU1FTc3d35fOcVlkxyL3ZMvr4lBfr1Sj3nwIED+eijj/Dw8CArK4spU6Ywb948evTogZ6ennq/0uofdujQgaZNm9K1a1fmzZtH27Ztgb/XeU2bNq1WrvMqNKzPgyajyDKzJ9f4VbLM7DWWpCYhIUEdQOzbt4927doRHx+vDjZ27979RKmN5ORkCgsL6dmzJzNmzOCnn36isLCQu3fv4uTkxOzZs0lPT0epVOLg4MCePXsAOHPmDJaWltSpU/rTwNJupFQX2jS2JbGwsuLNTo7YjxvNR1+tQ6X77K9jpdWsfViya9GiRbi7u+Pp6an+LJXL/m/h8WDzryeiQghREaQMi6hSMo4vTluy4Gr7WJa1tEVpQkNDWb58OQ0bNiy2PqosHBwc1FlrQfvKQ2j7WNYESUlJzJ07Vz0N9lHx8fEMHToUe3t7oqKisLOz47PPPuPHH39kwYIFFBQUYG9vz+LFizEyMlKX5dHX12fKlCnq6dWLFy/G3t4ePz8/0tPTUalU9O/fH39/f1JSUpg6dSq3b9/G2NiYpUuX8s9//pPAwEAUCoX66Wm3bt3YvHkzN27cYOHChejo6GBgYKA+t6gamvybLFg+G2JKuOHQsi160wKqvkPVmPzbWnPIWJZPaWVYJAAVVUrGsebQ9rGsqAB06NChTJw4kY4dO5b7HI8HoNrg0TVexo3/QU533xq3xquiPgNnzpzBwMBAvUazNEePHuXatWv4+/sXC+wmTZqkzj5bmbT9b1I8v7KMZUXf0Cr8OrBo+u1jdBxc0B09tdznrY3kb7LmkLEsn9ICUJmCK4So8W7duoWnpyeXLl2id+/eeHp64uPjQ2xsLADBwcHMnj1bvf/w4cM5c+YMQUFBnD9/nqlTp7JgwYJyv/65c+e0Lvh8dI1X9qmjssbrKc6ePcvFixefuZ+npyf+/v4sWrSo2Drbq1evcvToUbp16wYUfd5Gjx7N0KFDcXJyYuHChQBPlFVZt25d5VzQUygzCrh0VsmZ4+lcOqtEmVFQ5X0Qz+9hvVOTjCsYZv8Pk4wr1L27Ad3c++U/aZ+h8PjNqIZWRe1CCFEBJAAVQtRosbGxjBkzhqCgIFq0aMHevXs5evQo06ZNe2atvMmTJ6trcM6dO7eKelwFatEar7LU13RwcCA5uah25JUrV/D19SU+Pp6tW7fy1Vdf4eHhwblz57h//z5jxozB29sbb29vLly4APx9I8PHx6dYEpj4+HheeeUVoCgj8erVq4mIiCAvL49t27YREhJCQkJCsbIqx44dY9CgQVX6XikzCoj8LoOE23ncv1dAwu08Ir/LkCBUi1VGvVPdhlboTJ6PjoMLtGyLjoMLOpPn17gZEkIIzZHUmUKIGuv+/fuMGjWKr7/+Gjs7OxISEpg0aRJxcXHo6OiQl5f37JPUQKoHyWVqr87KU1/zUTY2NgwbNqzYOskJEyYwZswYOnbsSEJCAkOGDOHkyb+nLLZp04asrCzS09O5evUqhoaG1KtXD5VKxZw5c/j3v//Nr7/+SpcuXfj000/Vn007O7tiZVVcXFyq5D16KCY6m0xl8WU5mUoVMdHZvNFJUaV9Ec+nsuqd6ja0ApluK4SoJBKACiFqrDp16vDSSy9x/vx57OzsWLZsGZ07d2b9+vXEx8fj6+sLFK+rCcXrZNZEOnUtKWn1v05d7ZkmXFHKU1/zWb7//nuuXbum/j0jIwOlUllsnxYtWvDrr7+Sn5+PjY0NALm5ucTExPDFF1+QlZXFpUuXaNSoEQYGBuTn56vLqnz33Xds3bqV0NBQVqxY8aJvwXPLziosuT275Hahedpc71QIIUojAagQosYyNDRk/fr1DBkyBIVCQXp6OlZWRdPIduzYod7PxsaGzZs3U1hYSGJiYrH1ezVSn6Hwv5ji03Br6BqvkuprpqSklLjvozcinnYTorCwkNDQUIyNjUvdp2XLlhw/fpxr167RunVrdbudnR3/+te/iIqKIiCgKKPo8OHDgaKyKgYGBvTs2RNbW1s++OCD57vICmJsogs8Od3W2FhW62grpaUHBtnxxabhaku9UyGEKI38ryKEqNFMTU3ZvHkzX331Fa1bt2bx4sV4enqSn5+v3qe0Opk11eNrvIydPWvsGq+y1Ne0trYmKioKgIMHD6rPoVAoyMjIUP/u4uLCxo0b1b///PPPT7xugwYNyMnJwcrKChMTE6DohkhycjI3b94EIC8vr9ha0cTERHx9ffHw8OCDDz5g5syZFfEWPLeWbY0xVRQP2E0VOrRsW3qgLTRL2+udCiFESaQMi6hSMo41h4xlzVFTx7Ks9TXPnTvH1KlTqVOnDp06dSIqKopdu3Zx48YNxo0bh46ODgsXLqRFixbMmjWL2NhY8vPzcXBwYMmSJQQHB6ufbJZWhuXnn39m3rx5pKWlUVBQoM6Gm5HzB9H3dpGV9wATg7q0beiLmVGjMl1vRYyjMqOAmOhssrMLMTbWpWVbYxRmei90TlF2NfVvsraRcaw5ZCzLR+qACq0g41hz1NSxTE1NJTIykoyMDMzMzHB0dMTskY4BAAAgAElEQVTCwkLT3apUNXUsq4uMnD/47tYSlHl/qNsUBo3o2uzDMgWhMo41h4xlzSDjWHPIWJZPaQGorAEVQoi/pKamsnfvXtLS0tRtiYmJ9OvXr8YHoaJyPe3GRvS9XcWCTwBlXtET0U7W72uiu0IIIUSlkTWgQgjxl8jIyGLBJ0BaWhqRkZEa6lHNFh8fT7du3Z5onzZtWrEss9XdwxsbMTExJCQkEBMTw969e0lNTQUgK+9Bicdll9IuyiczM5Nhw4bh7u5Ot27d2L9/P1FRUQwYMIDu3bszZMgQfv/9d2JjY+nZs6f6uPj4eNzc3ABK3B/A19eXgIAAevbsSZcuXTh37pxGrlEIIaoDCUCFEOIvjyaaedTjJTZE5Vq+fDl2dnbPvX9BwZOZW7XJs25smBjULfE441LaRfmcOHECKysrIiIiOH78OK6ursyZM4d169Zx5MgRBg0axJIlS2jevDm5ubncvn0bgJCQEHr37k1eXl6J+z+Un5/PwYMH+fjjj6u0fI4QQlQ3EoAKIcRfzMzMSmxXKBRV3JPaIz8/H39/f1xcXBgzZgxZWVn4+vpy5coVAE6ePEnv3r3x8vJi7Nix6psBDg4OBAQE4OXlxYEDBzR5Cc/0rBsbbRv6ojAovtZTYdCItg19K71vtUmrVq04deoUAQEBnDt3jrt37xITE8M777yDh4cHn332GYmJiQD07t2bkJAQoCgA9fHx4caNG6XuD+Dt7Q1Au3btuHPnTtVfoBBCVBOyBlQIIf7i6OhIYmJisadV5ubm6jIdouLduHGDwMBAOnTowJQpU9i8ebN6W3JyMitXriQ4OBhTU1M+//xz1q1bx+TJkwGoV68eYWFhmur6c3vWjQ0zo6KEQ9H3dpGd9wDjcmbBFU9na2vLkSNHOH78OEuXLsXJyQk7OztCQ0Of2NfHx4dx48bRo0cPdHR0ePXVV/n1119L3R+KyuwA6OnpFSvzJIQQojgJQIUQ4i8WFhb069ePyMhIlEolCoWiVmTB1aQmTZrQoUMHAPr378+GDRvU2y5evMi1a9fo06cPUFQ3880331Rv9/HxqdrOltPz3NgwM2okCYcqWVJSEnXr1mXAgAGYm5uzZcsWkpOT+fHHH3nrrbfIy8vjf//7Hy1btuTll19GT0+PTz/9VP05s7W1LXV/IYQQz08CUCGEeISFhQVeXl6a7katoaOjU+rvKpUKZ2dn1qxZU+Kxpqamldq3iiI3NrTDb7/9xsKFC9HR0cHAwIDFixejp6f3RF3WhwGlj48PCxYsUK/VNTQ05Msvvyx1fyGEEM+nQuqADhw4cCqwHGi4Y8eO5ymSI3VAaykZx5pDxrLm0NRYxsfH4+joyP79+3nrrbeYNm0azZs3JyIigrlz52JtbU337t3ZsWMHr7zyCpmZmSQmJmJra4uDgwOHDx/G0tKyyvutreRvsuaQsawZZBxrDhnL8imtDugLJyEaOHCgDeAJ3H7RcwkhhKhdbG1t2bx5My4uLqSmpjJixAj1tvr16xMUFMSECRNwd3dXJ4IRQtskZeQSePouc8JvEXj6LkkZuZrukhBCaK2KmIIbBMwA9lfAuYQQQtQSNjY2nDp16on2Xbt2qX/u0qULhw4demIfqbMotEVSRi7/PRZPUkbeXy1ZXPszi4/dbLAyM9Ro34QQQhu90BPQgQMH9gESduzYcaWC+iOEEEI8ITU1lbCwMHbv3k1YWBipqama7pIQAHx75c9Hgs8iSRl5fHtFpusJIURJnvkEdODAgRGAVQmbZgOzKJp++0wDBw4cC4wF2LFjBw0aNChDN0FfX7/MxwjtI+NYc8hY1hzaPpbJycmEhoaSnJysbrt37x7Dhw+XNaCP0PZxrKky8kvOaZGRr1Pu8ZCxfNL48eOZOHEir732mqa78lwsLS1JS0sjNzeXKVOmsH37dq5cucLdu3fp0aOHprsnykj+JitWuZMQDRw4sC1wDMj8q8kauAt03LFjR9IzDpckRLWUjGPNIWNZc2j7WIaFhRETE/NEe8uWLSVj8SO0fRxrqsDTdzl1M+2JdueXzZnq1KRc55SxrP5atGhBSkpKsXEMDg4mKiqKgICA5z5Pfn4++vpStELT5G+yfCo8CdGOHTuid+zY0WjHjh0v79ix42XgDvDGcwSfQgghxHPLyMgosV2pVFZxT4R40lD7BliZGRRrszIzYKi9PC0pr8zMTIYNG4a7uzvdunVj//79+Pr6cuXKFe7cuYOTkxPJyckUFhbSr18/Tp48qekulyo+Pp5u3bqRm5vL8uXLCQkJwcPDg/3795OZmcmUKVPo2bMnnp6ehIWFAUWB6siRI/Hz82PQoEEavgIhKp7cUhFCCKHVzMzMSmxXKBRV3BMhnmRlZsjHbjZ8e+VPUrLyqGdi8FdQKgmIyuvEiRNYWVmxdetWANLS0tQ/W1tbM2HCBD766CNef/11WrRogYuLiya7+1wMDQ2ZNm1asSegixcvxsnJiRUrVpCamkrPnj15++23AYiOjiYiIoJ69eppsttCVIoKC0D/egoqhBBCVChHR0cSExNJS/t7mqO5uTmOjo4a7JUQf7MyMyz3dFvxpFatWjF//nwCAgJwd3fHwcGh2PYhQ4Zw4MABtm7dytGjRzXUyxd36tQpwsPDWbt2LQA5OTkkJCQA4OzsLMGnqLHkCagQQgitZmFhQb9+/YiMjESpVKJQKHB0dMTCwkLTXRNCVAJbW1uOHDnC8ePHWbp0KV26dCm2PSsri8TERKBoKn5psyS0nUqlYt26dTRv3rxY+6VLlzA1NdVQr4SofBKACiGE0HoWFhaScEiIWiIpKYm6desyYMAAzM3N2bZtW7HtAQEB9OvXD2tra6ZPn86WLVs01NOyMTMzK7am3cXFhY0bN7Jw4UJ0dHT4+eefadOmjQZ7KETVeKE6oEIIIYQQQlSk3377jV69euHh4UFQUBATJ05Ubzt79iyXL19mwoQJ9O/fH0NDQ4KDgzXY2+fXuXNnrv9/e3ceH0WV7n/8k31PIGyZCDiCgiwJypLADyUsCfuiCOXowGXUQVS4VxFEcWMQQRFlmQFFFFAYFAovqDEIhFWcGBa9hqBOkBgkbLKHpLMn/fsjkCESlATSlXS+79erX6ZPn6p+qp9ppp86p079+GPpIkRPPPEEBQUFREdH06NHD1577TWrQxRxiErfhuUa6TYstZTy6DyUS+ehXDoH5dF5KJfOQXl0Hspl5VzpNiyagisiIiIiNYJbXh4Bx3/BLb+QIk93MkMaUeTlZXVY1ywjI4PExESysrLw9/fXde7i1FSAioiIiEi155aXR3BqGh75BSUNNvCwZXOm+U01ugjNyMhg7dq1ZVb6PnbsGHfffbeKUHFKugZURERERKq9gOO//Kf4vMAjv4CA479YFNH1kZiYWKb4hJJ7nyYmJloUkUjVUgEqIiIiItWeW35hhdpriktXxr2UzWZzcCQijqECVERERESqvSLP8q8cu1J7TXGl+5j6+fk5OBIRx1ABKiIiIiLVXmZIIwo8Pcq0FXh6kBnSyKKIro/OnTsTGBhYpi0wMJDOnTtbFJFI1arZp4xEREREpFYo8vLiTPObnG4V3KCgIO6++24SExOx2Wz4+flpFVxxaipARURERKRGKPLy4tyNTa0O47oLCgqiT58+Voch4hCagisiIiIiIiIOoQJURESkGli/fj379+93yHvdcsstDnkfERGRX1MBKiIiUg38VgFaWFizbzMhIlJd6d9Xx9M1oCIiIlVkzpw5rFmzhnr16hEaGkp4eDh9+/blueee4/Tp0/j4+DBr1izOnj1LfHw8iYmJzJs3j3feeYeJEyfSunVrdu/ezZAhQ2jWrBl///vfyc/Pp27dusyfP58GDRqQnZ3N888/T2pqKjk5OUyYMIE+ffqwatUq4uPjycnJ4eDBg/Tr14/nn3++THxnzpxh1KhRPP7449x+++0888wzHDlyBICpU6fSqVMnKz42Eakl3nrrLTw9PXnooYeYMmUK33//PatXr+bLL79k5cqV+Pv7k5SURG5uLgMGDGDixIkAzJgxg40bN+Lu7k63bt148cUXiY2NZc6cObi6uhIYGMiaNWsoKipixowZfPXVV+Tn5zNq1ChGjhxJQkICs2bNIigoiAMHDvDll19a/EnULipARUREqsC3337LunXriI+Pp7CwkD59+hAeHs6kSZN49dVXadasGd988w2TJ09m9erVxMTEEB0dzcCBA0v3UVBQwOeffw7AuXPniI2NxcXFhQ8++IA333yTKVOmMG/ePLp27cqyZctITU1lwIAB3HnnnQB89913bNiwAU9PT7p168YDDzzADTfcAMDJkyd54IEHmDRpEt26dWPs2LGMHj2aiIgIjhw5wv3338/27dsd/8GJSK0RERHB22+/zUMPPcTevXvJz8+noKCAXbt2ERkZycCBA6lbty5FRUXce++9fP/994SEhPD555/zxRdf4OLiQkZGBgBz585lxYoV/OEPfyht+/DDDwkICGDdunXk5eVx1113ERUVBUBycjJbtmyhaVPnW9SqulMBKiIiUgV2795Nnz598Pb2BiAmJobc3Fy+/vprxowZU9ovPz//ivsYPHhw6d/Hjh3j0Ucf5cSJE+Tn55f+aPriiy+Ij4/n3XffpbCwkLy8vNJRzDvuuKP0/oItWrTgyJEj3HDDDRQWFnLvvfcyffp0unTpAsCOHTvKTAHOysoqvSWEiEhVCA8PJzk5mczMTDw9PQkLCyMpKYmdO3cybdo0YmNjWbFiBUVFRfzyyy/8+OOPtGjRAi8vLyZMmEB0dDTR0dEAdOzYkfHjxzNo0CD69esHwPbt2/nhhx+Ii4sDIDMzk7S0NDw8PLjttttUfFpEBaiIiIiD2O12AgMDiY+Pv6r+vr6+pX+/8MILPPzww/Tu3ZuEhARmz55dus9FixbRuXNnTp06Vdr/m2++wdPTs/S5q6tr6bVObm5uhIWFsW3bttICtLi4mNjY2NKCWUSkqnl4eNCkSRNM06Rjx460atWKhIQEDh48iLe3N2+//TZxcXHUqVOHJ554gtzcXNzd3YmLi+PLL78kLi6OpUuXsnr1ambOnMk333zD5s2b6devX+nskZdffpnu3buXed+EhIQy/76KY2kRIhERkSrQqVMn4uPjyc3NxWazsWnTJnx8fGjSpAmxsbFASfH43XffAeDv74/NZrvi/s6fP09ISAgAq1evLm2Piopi6dKl2O12APbt2/e7sbm4uDB79mxSU1NZsGBBmf1cdDX7ERG5VpGRkSxcuJDIyEgiIyNZvnw5bdu2JTMzEx8fHwIDAzl58iRbt24FwGazkZmZSa9evfjb3/7G999/D8DBgwdp3749Tz31FPXq1ePo0aNERUWxbNkyCgoKAEhNTSU7O9uyY5USGgEVERGpArfddhu9e/cmOjqaBg0a0KpVKwICApg/fz6TJ09m3rx5FBYWMmTIENq0acOQIUN46qmnWLx4MYsWLbpsfxMmTGDMmDEEBQXRtWtX0tPTAXjiiSeYMmUKHTp0oKCggCZNmrBs2bLfjc/NzY0FCxbwwAMP4Ofnx7Rp03j22WeJjo6msLCQyMhIZs6ced0/FxGRS0VERPD3v/+djh074uvri5eXFxEREbRp04a2bdvSrVs3QkNDSxdFy8rK4sEHHyQvLw+73c6UKVOAkpHOtLQ07HY7d9xxB23atKF169akp6fTt29f7HY7wcHBLFmyxMrDFcDl4hlTB7MfPXq0QhvUr1+/zNQiqZmUR+ehXDoP5bLqXLyGMicnh6FDh/Laa68RFhZWJe+lPDoP5dI5KI/OQ7msnNDQUACXX7drBFRERKSKTJo0if3795OXl8fw4cOrrPi8Vsez8lmRdIqz2QXU9fXgz+3qE+Lv+fsbiojUELasIlKSc8nNKcbbx5WWYd74+btZHVatpAJURESkily8vrI6O56Vz5TN6RzPKrjQksP+UzlM7dVERaiIOAVbVhGJ27LItl2c+VnE2dOFdO7uryLUAlqESEREpBZbkXTqkuKzxPGsAlYkabqZiDiHlOTcS4rPEtk2OynJuRZFVLupABUREanFzmYXlN+eU367iEhNk5tTXH57bvntUrVUgIqIiNRidX09ym/3Kb9dRKSm8fYpv+Tx9lYpZAV96iIiIrVYyYJDZYvNEP+ShYhERJxByzBvfP3KLsbq6+dCyzBviyKq3VSAiohIrbN48WKioqLo0KEDzz333HXb77Bhw0hKSrpu+3OEEH9PpvZqQrc/BhLWyIdufwzUAkQi4lT8/N3o3N2fG5p6UK+hGzc09dACRBbSKrgiIlLrvP/++6xcuZIdO3awd+9eq8OxXIi/JxO6hlodhohIlfHzd6N9Fz+rwxA0AioiIrXM008/zaFDhxg5ciQZGRml7Rs3bmTgwIH07t2be++9l5MnTwLw1VdfERMTQ0xMDL179yYrKwsoucVKr169iI6OZsaMGaX7+eyzzxgwYAB33HEHO3fuBKCoqIhp06bRv39/oqOjWb58uQOPWEREpPrQCKiIiNQqM2fOZNu2baxevZr4+PjS9oiICGJjY3FxceGDDz7gzTffZMqUKSxcuJAZM2bQqVMnbDYbXl5ebNmyhQ0bNvDZZ5/h4+PD2bNnS/dTWFhIXFwcmzdvZvbs2axatYoPP/yQgIAA1q1bR15eHnfddRdRUVE0bdrUio9ARETEMipARUREgGPHjvHoo49y4sQJ8vPzS4vDTp06MXXqVO6++2769etHaGgoO3bs4N5778XHxweAunXrlu6nf//+AISHh3P48GEAtm/fzg8//EBcXBwAmZmZpKWlqQAVEZFaRwWoiIgI8MILL/Dwww/Tu3dvEhISmD17NgDjxo2jV69ebNmyhbvuuosPPvjgN/fj6VmyeI+bmxuFhYWl7S+//DLdu3evsvhFRERqAl0DKiIiApw/f56QkBAAVq9eXdp+8OBBWrVqxdixY2nXrh0HDhygW7durFq1ipycHIAyU3DLExUVxbJlyygoKAAgNTWV7OzsKjoSERGR6ksjoCIiIsCECRMYM2YMQUFBdO3alfT0dADeffddEhIScHV1pUWLFvTo0QMvLy++++47+vXrh4eHBz179mTy5MlX3Pf9999Peno6ffv2xW63ExwczJIlSxx1aCIiItWGi91ut+J97UePHq3QBvXr1+fUqVNVFI44ivLoPJRL56FcOgfl0Xkol85BeXQeymXlhIaGArj8ul0joCIiIlXIllVESnIuuTnFePu40jLMWzc/FxGRWksFqIiISBWxZRWRuC2LbNvF2UZFnD1dSOfu/ipCRUSkVtIiRCIiIlUkJTn3kuKzRLbNTkpyrkURiYiIWEsFqIjIFSxbtqzMaqhXem3VqlUcP37ckaFJDZGbU1x+e2757SIiIs5OU3BFRK7gv/7rv8ptLywsLPPa6tWrufXWW0tv4SFykbePK1B0ebu3zv+KiEjtpAJURGqV7OxsxowZw7FjxyguLubxxx9nxowZDBo0iK1bt+Lt7c38+fO56aabeOONN/Dz8+ORRx5h2LBhtG7dmt27dzNkyBBsNht+fn40btyYpKQkxo0bh7e3N59++indu3fn888/Jzg4mKSkJKZNm8ZHH33E2bNnmTBhAocOHcLb25vXXnuN1q1b89VXX/Hiiy8C4OLiwpo1a/D39+ett94iNjaW/Px8+vbty8SJE0lPT2fEiBFERESwZ88eQkJCWLJkCT4+PhZ/slKelmHenD1dWGYarq+fCy3DvC2MSkRExDo6BSsitcrWrVsJCQlh06ZNbNmyhR49egAQEBDA5s2b+ctf/sKUKVPK3bagoIDPP/+cRx55pLRt4MCBtGvXjvnz5xMfH/+bheAbb7xB27Zt2bRpE8888wyPP/44AAsXLmTGjBnEx8ezdu1avL292b59O2lpacTFxbFx40b27t1LYmIiAGlpaYwaNYqtW7cSGBjIunXrrtfHI9eZn78bnbv7c0NTD+o1dOOGph5agEhERGo1FaAiUqvceuutfPHFF0yfPp2dO3cSGBgIwF133VX636+//rrcbQcPHnxN771r1y7uueceAO644w7Onj1LZmYmnTp1YurUqSxevJiMjAzc3d3Zvn0727dvp3fv3vTp04fU1FTS0tIAaNKkCW3btgUgPDyc9PT0a4pLqpafvxvtu/jx/3oE0L6Ln4pPERGp1TQFV0RqlebNm7N+/Xq2bNnCa6+9xh133AGUTH296NK/L+Xr63tV7+Hu7k5xcckiM3l5eb/bf9y4cfTq1YstW7Zw11138cEHH2C32xk3bhwjR44s0zc9PR0vL6/S525ubuTmakVVERERqRk0Aioitcrx48fx8fHhnnvu4ZFHHiE5ORmATz/9tPS/HTp0qNA+/fz8yMrKKn3euHFj9u7dC0BcXFxpe2RkJGvWrAEgISGB4OBgAgICOHjwIK1atWLs2LG0a9eOAwcO0L17d1atWoXNZgPg2LFjnDp1qvIHLiIiIlINaARURGqVf//737z88su4uLjg4eHBK6+8wsMPP0xGRgbR0dF4enqyYMGCCu3TMAyeeeaZ0kWInnzySSZMmMCsWbPo0qVLab+L7dHR0Xh7ezN37lwA3n33XRISEnB1daVFixb06NEDLy8vfvzxx9Jpv76+vvzjH//AzU3TN0VERKTmcrHb7b/f6/qzHz16tEIb1K9fX2f/nYDy6DycKZeRkZGlq9bWRs6Uy9pMeXQeyqVzUB6dh3JZOaGhoQCXXdekKbgiIg4WGRnJmTNnSE9Pp2fPnle93bBhw0jevZs6Px+i3o8/UefnQ7j9xjWmCQkJpfcr3bhxI/Pnz690rCIiIiLXg6bgikitt3PnTqtDuDrFxQQdPoqvx4V7SNrAw5bNmeY3UXTJwkTl6d27N71793ZAkCIiIiJXphFQEZEq9OCDD9K3b1969OjBP//5zyv2+/nnn+nduzfffvst+/btY+DAgURHR/PQQw9x7tw5ANzz81m7dSsR4/+HFqMfYse+fXjkF+Dx8yHGjx9Pr1696N27N//6178u2/+qVat47rnnAIiNjaVnz55ER0czdOhQAIqKinjppZdK25csWVK67ZIlS+jTpw+9evXiwIEDAGRnZ/Pkk08yYMAAevfuzYYNG67bZyYiIiLOSyOgIiJV6I033qBu3brk5OQwYMAA+vfvf1mfAwcO8NhjjzFnzhzatGlDdHQ006ZNo0uXLsyaNYvZs2fz0ksvQbGdwuIids35O+t272LqB/9k04xXecdcjYuLC5s3b+bAgQPcd9997Nix44oxzZ07lxUrVvCHP/yBjIwMAP75z3+Snp7Oxo0bcXd35+zZs6X9g4OD2bBhA++99x4LFy7k9ddfZ968eXTt2pXZs2eTkZHBgAEDuPPOO6/6VjUiIiJSO2kEVESkCi1ZsoTo6GgGDRrE0aNHSUtLK/P66dOnefDBB5k/fz5t2rTh/PnzZGRklK6eO3z48P9MEXZ1Yej/6wpAh5tv4eCJXwD413fJpSOZN998M40bN+ann366YkwdO3Zk/PjxrFixgqKiIgC+/PJLRo4cibt7yXnJunXrlvbv168fAOHh4aSnpwPwxRdfsGDBAmJiYhg2bBh5eXkcOXLkmj4rERERcX4aARURqSIJCQns2LGD2NhYfHx8Sgu1SwUEBHDDDTewa9cuWrRo8Zv7K/T0xM3HBwA3V1cKi4oo8PSg0Nu7QnHNnDmTb775hs2bN9OvXz927dr1m/29Llxf6ubmVlqw2u12Fi1axM0331yh9xYREZHaTSOgIiJVJDMzk6CgIHx8fDhw4ADffPPNZX08PT1ZvHgxH330EWvXriUwMJCgoKDSUc///d//pXPnziWdXV3JaBxKdt0g8nx9sbu6cqb5TXTq3Jm1a9cCkJqaypEjR2jevPkV4zp48CDt27fnqaeeol69ehw+fJg777yT5cuXU1hYCFBmCm55oqKiWLp0KRdv5bVv374Kfz4iIiJS+2gEVESkinTv3p3ly5cTFRVF8+bNad++fbn9fH19ef/997nvvvvw8/Nj7ty5PPPMM+Tm5tK0aVNmz55d2rfY05NzNzblbIA/xe7uFHl5MWrUKCZPnkyvXr1wc3Njzpw5paOW5Xn55ZdJS0vDbrdzxx13EB4eTqNGjfjpp5+Ijo7G3d2dP//5zzzwwANX3McTTzzBlClTiI6Opri4mCZNmrBs2bLKf1giFsvOzmbMmDEcO3aM4uJiHn/8cW666SamTp2KzWYjODiYOXPmkJmZyeOPP05cXBwA6enp/OUvf2Hz5s3s3bv3sv6NGjVi2LBh3H777SQkJJCRkcEbb7xBZGSkxUcsImINl4tnrx3MfvTo0QptoBvAOgfl0Xkol85DuXQOyuO1iYuLY9u2bcyaNQuA8+fPM2LECJYuXUq9evX45JNP2L59O7NnzyYmJobFixfTtGlTFixYQEFBAWPHjuWee+4pt/+wYcMICwtjypQpbN68mUWLFrFq1aorxqJcOgfl0Xkol5UTGhoK4PLrdo2AiohIhWTlnSD55EfkFJzDx6MOYQ2G4e/V0OqwRK7JrbfeyksvvcT06dOJjo4mKCiIlJQU/vSnPwFQXFxMw4Yl/zsfNGgQn376KePGjePTTz/lrbfeIjU19Yr9gdIVsMPDwzl8+LCDj05EpPpQASoiIlctK+8E236eia3gRGnb6exUut/4tIpQqdGaN2/O+vXr2bJlC6+99hpdu3alRYsWxMbGXtZ38ODBjBkzhn79+uHi4kKzZs344YcfrtgfSq73hpLFvC5eay0iUhtpESIREblqySc/KlN8AtgKSkZERWqy48eP4+Pjwz333MMjjzzC//3f/3HmzBn27NkDQEFBASkpKQD88Y9/xM3Njblz5zJ48GCgpIC9Un8REfkPjYCKiKWKA0oAABt2SURBVMhVyyk4V2577hXaRWqKf//737z88su4uLjg4eHBK6+8gpubGy+++CLnz5+nqKiIv/71r7Rs2RIoGQWdNm0aiYmJQMkI59tvv33F/iIiUkKLEIlDKY/OQ7l0HhXJ5VeH3+RQxleXtTcN6kKXxo9d79CkAvSddB7KpXNQHp2Hclk5WoRIRESuWViDYZzOTi0zDdfPoyFhDYZZGJVI9VV88jh8sgL7uTO41AmGIX/GtUGI1WGJiFhGBaiIiFw1f6+GdL/xaZJPfkRuwTm8tQquyBUVnzyOfc6LcPI4AHaAn1IoHv9SjS1C33777dJbyNx333307duXESNGEBERwZ49ewgJCWHJkiX4+Phw8OBBnnvuOU6fPo2Pjw+zZs3i5ptvtvgIRMRqWoRIREQqxN+rIV0aP0aPm56lS+PHVHyKXMknK0qLz1IXRkRror1792KaJp999hmxsbF88MEHZGRkkJaWxqhRo9i6dSuBgYGsW7cOgEmTJjFt2jTWr1/PCy+8wOTJky0+AhGpDjQCKiIiIlIF7OfOVKi9utu1axd9+/bF19cXgH79+rFz506aNGlC27ZtgZL7nKanp2Oz2fj6668ZM2ZM6fb5+fmWxC0i1YsKUBEREZEq4FInmPKWenSpE+zwWKqSl5dX6d9ubm7k5uZSXFxMYGAg8fHxFkYmItWRpuCKiIiIVIUhf4ZfX+vZIKSkvQaKjIxkw4YN5OTkkJ2dzfr164mMjCy3b0BAAE2aNCE2NhYAu93Od99958hwRaSa0gioiIiISBVwbRBC8fiXnGYV3LCwMIYPH86AAQOAkkWIgoKCrth//vz5TJ48mXnz5lFYWMiQIUNo06aNo8IVkWpK9wEVh1IenYdy6TyUS+egPDoP5dI5KI/OQ7msHN0HVEREREQc4nhWPiuSTnE2u4C6vh78uV19Qvw9rQ5LRKoBFaAiIiIict0cz8pnyuZ0jmcVXGjJYf+pHKb2aqIiVES0CJGIiIiIXD8rkk5dUnyWOJ5VwIokTWEUERWgIiIiInIdnc0uKL89p/x2EaldVICKiEiNMnHiRPbv3291GCJyBXV9Pcpv9ym/XURqFxWgIiJSo7z++uu0aNHC6jBEaqWMjAzee++93+xTsuBQ2WIzxL9kISIRERWgIiJiibfeeovFixcDMGXKFIYPHw7Al19+ybhx49i+fTuDBg2iT58+PPzww9hsNgCGDRtGUlISALfccguvvvoq0dHRDBw4kJMnTwJw+vRpRo8eTf/+/enfvz+7d++24AhFnM/58+dZtmzZb/YJ8fdkaq8mdPtjIGGNfOj2x0AtQCQipVSAioiIJSIiIti5cycAe/fuJTs7m4KCAnbt2kWrVq2YN28eq1atYsOGDbRr145FixZdto/s7Gzat2/Ppk2b6Ny5MytWrADgxRdfZPTo0axbt4533nmHiRMnOvTYRJzVjBkz+Pnnn4mJiWH8+PFs3LgRgIceeognn3wSgJUrV/Le/NlM6BpKk9T1xL0wgvsH9+Wdd96xMnQRqSZ0GxYREbFEeHg4ycnJZGZm4unpSVhYGElJSezcuZPevXuzf/9+hgwZAkBBQQEdOnS4bB+enp7ExMQAEBYWxo4dOwDYsWNHmetEs7KysNls+Pn5OeDIRJzXs88+S0pKCvHx8XzyySel39fjx4/zyy+/ALBz506GDBnC3r17MU2Tzz77DLvdzsCBA+nSpQtt27a1+ChExEoqQEVExBIeHh40adIE0zTp2LEjrVq1IiEhgYMHD9KkSRO6devGm2+++Zv7cHd3x8XFBQA3NzcKCwsBKC4uJjY2Fm9v7yo/DpHaKiIignfeeYf9+/dzyy23kJGRwS+//MLXX3/NtGnTWLlyJX379sXX1xeAfv36sXPnThWgIrWcpuCKiIhlIiMjWbhwIZGRkURGRrJ8+XLatm1Lhw4d2L17N2lpaUDJVNvU1NSr3m9UVBRLly4tfb5v377rHrtIbfeHP/yB8+fPs3XrVjp37kxkZCSxsbH4+fnh7+9/WX+73W5BlCJS3agAFRERy0RERHDixAk6duxIgwYN8PLyIiIignr16jFnzhzGjh1LdHQ0gwcPrlABOm3aNJKSkoiOjqZ79+4sX768Co9CxHktXryYqKgoWrduzfz58/Hz8+OXX34hJiaGmJgYjh07xquvvoqLiwumaTJt2jQyMzOx2WxERkayYMEC/va3vxETE8Pq1avJzs6mV69e9OzZk+nTp1t9eCJiAReLzkbZjx49WqEN6tevz6lTp6ooHHEU5dF5KJfOwxlz6Zp/Gr8z8bgVZlLkHoAtOIZiz3pWh1WlnDGPtVV1ymW3bt1YuXIloaGhpW1jx47lhx9+ICoqivj4eE6fPk2rVq1YunQpHTp0oF+/fjRr1ozx48fTunVrPD09CQ4OZuDAgXz44YesX7+eoKAg7rvvPh566CH69u1r4RFWneqUR7k2ymXlXPh3w+XX7RoBFRERp+Kaf5o6R5fgk5WEZ+5P+GQlUefoElzzT1sdmkiN8vTTT3Po0CFGjhzJokWLeO655zh//jy7d+9m06ZN5ObmMmDAANzd3UlJSWHw4MG4uroSFxfHe++9x4EDBwgICOCzzz5jy5YttGnThi5dulCvXj3c3d0ZOnQoiYmJVh+miDiYClAREXEqfmficS88U6bNvfAMfmfiLYpIpGaaOXMmjRo1YvXq1QQFBQEQGBhImzZtmDFjBocPH6ZVq1a0adOGqKgoGjVqxPr16/npp59YunQpT0+aTH6enX1fF/PNVzZyc4otPiIRqQ60Cq6IiDgVt8LMCrWLSMXcdtttvP322yQkJDBhwgRGjBjBlClTOHPmDGPGjKG4uJj8/AJysu0UFto5e6qIgtwCCgtv4auvSvoFBQXx8ccf8+CDD1p9OCLiYBoBFRERp1LkHlChdhGpmB9//JHMzEyGDh3Kpk2b2LJlC6+88grFxcXY7XZcXFy4f/hEpkyKLbOdp3t97jcmMHz4cGJiYggPD6dPnz4WHYWIWEUjoCIi4lRswTF45KaXmYZb6B6MLTjGwqhEnMf8+fPJz8/Hy8uLyMhIXnnlFQDatWvH6NGjGTRoEP/afJ69Sd8z44WyU98jO/Zn/FP3WhF2lZg1axaRkZF069bN6lBEagwVoCIi4lSKPetxLvTBWrcKrogjDR48mDFjxvDRRx+Vts2fP5/Jkyczb948sjLzua1NXxrfcGuZ7bwOp1D0+lpc6gTDkD/j2iDE0aFfV0899ZTVIYjUOLoNiziU8ug8lEvnoVw6B+XReThDLm1ZRSRuyyLb9p/fmT55p4jcPQPf3AvH1iAEl/Ev1YgiND09nREjRhAREcGePXsICQlhyZIlTJ48mejoaAYOHEhkZCTDhw8nPj6ewsJCVq1aRf369cnOzub5558nJSWFgoICJkyYoKnHNYwzfCetoNuwiIiIiEiVyMo7wVeH32RL2gy+Ovwmdo/TdO7uzw1NPajX0I3Q/ANli0+Ak8fhkxXWBV1BaWlpjBo1iq1btxIYGMi6desu6xMcHMyGDRsYOXIkc+fOBWDevHl07dqVuLg4Vq9ezbRp08jOznZ0+CLVhqbgioiIiEilZeWdYNvPM7EVnChtO52dSvcbn6Z9l4YAFL2+FnIvH0GynztzWVt11aRJE9q2bQtAeHg46enpl/Xp169f6eubNm0C4IsvviA+Pp6FCxcCkJeXx5EjR7jlllscFLlI9aICVEREREQqLfnkR2WKTwBbwQmST35El8aPAeBSJ5jyLvpyqRPsgAivDy8vr9K/3dzcyM3NvWIfNzc3ioqKALDb7SxatIibb77ZMYGKVHOagisiIiIilZZTcK7c9txL24f8GX59rWeDkJJ2JxcVFcXSpUu5uO7Kvn37LI5IxFoqQEVERESk0nw86pTb7n1Ju+uFBYdcIqOgZRgukVE1ZgGia/XEE09QUFBAdHQ0PXr04LXXXrM6JBFLaRVccSjl0Xkol85DuXQOyqPzqGm5LO8aUD+PhnS/8Wn8vRpaGJm1vDwD+Wr7MXJzivH2caVlmDd+/m5WhyWVUNO+k9XFlVbB1TWgIiIiIlJp/l4lxWbyyY/ILTiHt0cdwhoMq9XFpy2riG07jpJ5vvBCSxFnTxfSubu/ilCp9VSAioiIiMg18fdqWLrgkEBKcu4lxWeJbJudlORc2nfxsygqkepB14CKiIiIiFxHuTnF5bfnlt8uUpuoABURERERuY68fcr/ie3trZ/eIvoWiIiIiIhcRy3DvAkILHulm6+fCy3DvC2KSKT60DWgIiIiIiLXkZ+/G32GhJasgptbjLe3VsEVuUgFqIiIiIjIdRYQ6KkFh0TKoSm4IiIiIiIi4hAqQEVERERERMQhVICKiIiIiIiIQ1zzNaCGYfw3MBYoAuJM05x0zVGJiIiIiIiI07mmEVDDMHoAQ4B2pmm2AV6/LlGJiIiIiIiI07nWKbiPAq+appkHYJrmiWsPSURERERERJzRtU7BbQHcaRjGdCAXmGia5u5rD0tEREREREScjYvdbv/NDoZhbAJCynnpOWA6sBX4H6ATsApoZprmZTs1DONh4GEA0zQ75OfnVyhQd3d3CgsLK7SNVD/Ko/NQLp2HcukclEfnURNyee7cOVauXMkjjzzC9u3bmTNnDh9//LHVYVUrNSGPcnWUy8rx9PQEcPl1+++OgJqmGX2l1wzDeBRYc6Hg3GUYRjFQHzhZzn4WAYsuPLWfOnXq6iK/oH79+lR0G6l+lEfnoVw6D+XSOSiPzqMm5DI9PZ0333yTYcOGkZGRQX5+/u/GXFRUhJubm4MitF5NyKNcHeWyckJDQ8ttv9YpuB8DPYCthmG0ADwBZUdERETEic2YMYOff/6ZmJgYPDw88PX1ZfTo0aSkpBAeHs4//vEPXFxciIyMZPDgwXzxxRc89thj1KlTh9dff538/HxuvPFG5syZg5+fH3v37mXq1KnYbDaCg4OZM2cOjRo1svowRaQKXOsiREuAZoZh7ANWAqPKm34rIiIiIs7j2Wef5cYbbyQ+Pp7nn3+effv2MXXqVLZt28bPP//M7t3/WRKkbt26bNiwgTvvvJN58+axatUqNmzYQLt27Vi0aBEFBQU8//zzLFq0iPXr13Pvvfcyc+ZMC49ORKrSNY2AmqaZD4y4TrGIiIiISA102223lU63a9OmDenp6URERAAwePBgAL7++mv279/PkCFDACgoKKBDhw6kpqaSkpLCn/70JwCKi4tp2LChBUchIo5wrVNwRURERKSWu7DYCABubm5lFmzx9fUFwG63061bN958880y2/7www+0aNGC2NhYxwQrIpa61im4IiIiIlLL+Pn5kZWVVaFtOnTowO7du0lLSwMgOzub1NRUmjdvzpkzZ9izZw9QMjKakpJy3WMWkepBI6AiIiIiUiHBwcF06tSJnj174u3tTf369X93m3r16jFnzhzGjh3LxdvxTZo0iebNm/P222/z4osvcv78eYqKivjrX/9Ky5Ytq/owRMQCv3sf0CpiP3r0aIU20PLHzkF5dB7KpfNQLp2D8ug8lEvnoDw6D+Wyci5cF17x+4CKiIiIiFSVjIwMEhMTycrKwt/fn86dOxMUFGR1WCJSRXQNqIiIiIhUucWLFxMVFcW4ceNK2zIyMli7di0pKSkcOXKElJQU1q5dS0ZGhoWRikhV0gioiIiIiFS5999/n5UrV5bergUgMTGR8+fPl+l3/vx5EhMT6dOnz2X7sNvt2O12XF01hiJSU6kAFREREZEq9fTTT3Po0CFGjhzJ8OHD2bVrF4cOHSI7O5uePXvSsGFDvvzySzw8PIiMjMRms9GzZ0/ef/99AO6//35uv/12kpOTWb58OY0bN7b4iESksnT6SERERESq1MyZM2nUqBGrV6/m8OHDtG3blk2bNmEYBnFxcZf19/PzK/M8LS2NUaNGsXXrVhWfIjWcClARERERcZhdu3Zxzz33APDAAw+Ql5dHXl5e6euBgYF07ty5zDaNGzemQ4cODo1TRKqGClARERERsURQUBA+Pj7ccsst1KlThwYNGnD33XcTFBRUpij19fW1MEoRuZ5UgIqIiIiIw0RGRrJmzRoAEhISqFevHnfddRd9+vQhLy+PoKAgkpOTOXTokMWRikhVUAEqIiIiZaSnp9OzZ8/L2idOnMj+/fstiEicyZNPPklycjLR0dHMmDGDuXPnAtC/f3/OnTtHjx49WLp0Kc2aNbM4UhGpCloFV0RERK7K66+/bnUIUoPt3Lmz9O8lS5Zc9rqPjw8ffvhhudtu2bKlyuISEcfSCKiIiIhcprCwkHHjxhEVFcXo0aPJyclh2LBhJCUlAfDxxx/Tq1cvevbsyfTp0y2OVpyRa/5pAo6vpM7hdwg4vhLX/NNWhyQi14EKUBEREblMamoqo0aNYvv27QQEBJTejxHg+PHjTJ8+HdM02bhxI99++y3r16+3MFpxNq75p6lzdAk+WUl45v6ET1YSdY4uUREq4gRUgIqIiMhlQkND6dSpEwBDhw5l165dpa8lJSXRpUsX6tWrh7u7O0OHDiUxMdGqUMUJ+Z2Jx73wTJk298Iz+J2JtygiEbleVICKiIjIZVxcXH7zuUhVcivMrFC7iNQcKkBFRETkMkeOHGHPnj1AyfWeF0dDAW677TYSExM5c+YMRUVFfPzxx3Tp0sWqUMUJFbkHVKhdRGoOFaAiIiJymebNm/P+++8TFRVFRkYGo0aNKn2tUaNGPPvsswwfPpyYmBjCw8Pp06ePhdGKs7EFx1DoHlymrdA9GFtwjEURicj14mK32614X/vRo0crtEH9+vU5depUFYUjjqI8Og/l0nkol85BeXQeymUJ1/zT+J2Jx60wkyL3AGzBMRR71rM6rKumPDoP5bJyQkNDAS67fkP3ARUREZEKsWUVkZKcS25OMd4+rrQM88bP383qsMTJFHvWIzPkT1aHISLXmQpQERERuWq2rCISt2WRbbs4g6qIs6cL6dzdX0WoiIj8Ll0DKiIiIlctJTn3kuKzRLbNTkpyrkURiYhITaICVERERK5abk5x+e255beLiIhcSgWoiIiIXDVvn/J/Onh76yeFiIj8Pv2/hYiIiFy1lmHe+PqVXdTQ18+FlmHeFkUkIiI1iRYhEhERkavm5+9G5+7+Javg5hbj7a1VcEVE5OqpABUREZEK8fN3o30XP6vDEBGRGkhTcEVERERERMQhVICKiIiIiIiIQ6gAFREREREREYdQASoiIiIiIiIOoQJUREREREREHEIFqIiIiIiIiDiEClARERERERFxCBWgIiIiIiIi4hAqQEVERERERMQhVICKiIiIiIiIQ6gAFREREREREYdQASoiIiIiIiIOoQJUREREREREHEIFqIiIiIiIiDiEClARERERERFxCBWgIiIiIiIi4hAqQEVERERERMQhVICKiIiIiIiIQ6gAFREREREREYdQASoiIiIiIiIOoQJUREREREREHEIFqIiIiIiIiDiEClARERERERFxCBWgIiIiIiIi4hAqQEVERERERMQhVICKiIiIiIiIQ6gAFREREREREYdQASoiIiIiIiIOoQJUREREREREHEIFqIiIiIiIiDiEClARERERERFxCBWgIiIiIiIi4hAqQEVERERERMQhVICKiIiIiIiIQ6gAFREREREREYdwsdvtVryvJW8qIiIiIiIiDuPy6warRkBdKvowDOPrymynR/V6KI/O81AuneehXDrHQ3l0nody6RwP5dF5HsrlNT0uoym4IiIiIiIi4hAqQEVERERERMQhalIBusjqAOS6UB6dh3LpPJRL56A8Og/l0jkoj85DubyOrFqESERERERERGqZmjQCKiIiIiIiIjWYu9UBVJRhGP8NjAWKgDjTNCdZHJJUkmEYE4DXgQamaZ6yOh6pOMMwZgGDgHwgFXjANM1z1kYlV8swjL7APMANeNc0zVctDkkqwTCMJsAyoBEltzlbZJrmPGujksoyDMMN2AMcMU1zoNXxSOUYhlEHeBdoS8n38kHTNL+yNiqpKMMwxgN/pSSHyZT8zsm1Nqqar0aNgBqG0QMYArQzTbMNJcWL1EAXfjD1Bg5ZHYtck3igrWma4cB+YLLF8chVuvAjdwHQD2gN3GcYRmtro5JKKgQmmKbZGugMjFUua7THgR+sDkKu2TxgvWmatwLtUE5rHMMwbgD+B+hommZbSk7W/snaqJxDTRsBfRR41TTNPADTNE9YHI9U3hxgEvCJ1YFI5ZmmufGSp4nAMKtikQqLAA6YpvkTgGEYKyk5wfe9pVFJhZmmeQw4duHvTMMwfgBuQLmscQzDaAwMAKYDT1ocjlSSYRhBQDfgLwCmaeZTMlNIah53wMcwjALAFzhqcTxOoUaNgAItgDsNw9hpGMZ2wzA6WR2QVJxhGEMomVqUZHUscl09CHxudRBy1W4A0i95fvhCm9RghmH8Ebgd2GlxKFI5cyk5OVtsdSByTW4CTgJLDcP4P8Mw3jUMw8/qoKRiTNM8Qslsy0OUnOTL+NWJd6mkajcCahjGJiCknJeeoyTeYEqmGHUCTMMwmpmmqaV8q5nfyeOzlEy/lRrgt3JpmuYnF/o8R8k0wBWOjE1E/sMwDH/gf4EnTNM8b3U8UjGGYQwETpim+bVhGN2tjkeuiTvQHvhv0zR3GoYxD3gGeMHasKQiDMOoS8nMoJuAc8BqwzBGmKb5T2sjq/mqXQFqmmb0lV4zDONRYM2FgnOXYRjFQH1KzjJJNXKlPBqGEUbJFznJMAyAxsA3hmFEmKZ53IEhylX6re8kgGEYfwEGAr10MqhGOQI0ueR54wttUgMZhuFBSfG5wjTNNVbHI5XSFRhsGEZ/wBsINAzjn6ZpjrA4Lqm4w8Bh0zQvzkT4iJICVGqWaCDNNM2TAIZhrAH+H6AC9BpVuwL0d3wM9AC2GobRAvAEtHpqDWKaZjLQ8OJzwzAOUnJxt/JYA11YRXUSEGWaZrbV8UiF7AZuMQzjJkoKzz8B91sbklSGYRguwGLgB9M0Z1sdj1SOaZqTubCQ24UR0IkqPmsm0zSPG4aRbhhGS9M0U4Be6JrsmugQ0NkwDF8gh5I87rE2JOdQ064BXQI0MwxjH7ASGKURFxFLzQcCgHjDML41DGOh1QHJ1TFNsxAYB2ygZHVG0zTN76yNSiqpKzAS6Hnhe/jthVE0EbHOfwMrDMPYC9wGzLA4HqmgCyPYHwHfUHILFldgkaVBOQkXu131m4iIiIiIiFS9mjYCKiIiIiIiIjWUClARERERERFxCBWgIiIiIiIi4hAqQEVERERERMQhVICKiIiIiIiIQ6gAFREREREREYdQASoiIiIiIiIOoQJUREREREREHOL/A+FN3A6K96ToAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 1152x1152 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vvW651_xY3Pd",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 918
        },
        "outputId": "d3641817-f3f3-488f-d8e0-80bb91698eec"
      },
      "source": [
        "tsne_plot(de_muse,fr_muse,fr_muse_align)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1152x1152 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j6Jqaz6RTouh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "995b68b7-402a-424f-e828-bd226c02788f"
      },
      "source": [
        " words_de =  [repla(w) for w in de_fr_dict.keys()]\n",
        "words_fr =  [repla(w) for w in de_fr_dict.values()]\n",
        "embed_de =  [de_muse[w] for w in words_de]\n",
        "embed_fr =  [fr_muse[w] for w in words_fr]\n",
        "words = words_de + words_fr\n",
        "embed = embed_de+embed_fr\n",
        "embed[:2]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[array([-5.500e-03,  6.590e-02,  8.730e-02, -4.080e-02,  2.600e-03,\n",
              "         4.410e-02, -1.100e-02, -1.007e-01,  1.202e-01, -6.500e-02,\n",
              "        -2.230e-02,  2.600e-03,  1.156e-01, -1.600e-03, -1.680e-02,\n",
              "        -1.650e-02,  6.360e-02, -7.040e-02, -2.560e-02, -9.600e-02,\n",
              "         8.300e-02,  4.040e-02, -1.100e-02,  5.040e-02,  5.020e-02,\n",
              "         6.260e-02,  1.730e-02, -7.650e-02,  3.750e-02,  2.580e-02,\n",
              "         2.210e-02, -3.340e-02,  1.860e-02,  1.075e-01,  5.310e-02,\n",
              "        -1.070e-02,  1.530e-02, -2.410e-02,  4.870e-02,  6.800e-03,\n",
              "         3.060e-02,  5.210e-02,  3.000e-02,  3.440e-02, -6.850e-02,\n",
              "         7.390e-02,  5.840e-02, -5.630e-02,  3.200e-03, -4.230e-02,\n",
              "         2.250e-02,  8.000e-04,  5.580e-02, -5.950e-02, -4.000e-04,\n",
              "        -2.670e-02,  1.780e-02,  1.310e-02, -1.930e-02, -8.120e-02,\n",
              "        -1.041e-01, -1.650e-01,  1.084e-01, -6.180e-02,  3.690e-02,\n",
              "        -1.051e-01, -1.199e-01, -6.800e-02, -5.040e-02, -4.770e-02,\n",
              "        -1.429e-01,  2.100e-03,  1.290e-02, -1.517e-01, -6.160e-02,\n",
              "        -8.610e-02, -3.600e-02, -2.190e-02, -8.300e-03, -6.860e-02,\n",
              "         4.250e-02, -7.910e-02, -6.100e-03,  1.800e-03,  7.330e-02,\n",
              "         6.660e-02,  4.960e-02,  6.490e-02,  9.070e-02,  8.140e-02,\n",
              "        -6.600e-03,  1.234e-01, -6.680e-02, -6.350e-02, -4.080e-02,\n",
              "         5.380e-02,  9.200e-03, -5.640e-02,  5.300e-02,  8.400e-02,\n",
              "        -1.328e-01, -5.450e-02, -6.320e-02, -5.550e-02, -8.450e-02,\n",
              "         7.410e-02, -1.400e-03,  2.570e-02, -1.332e-01,  1.010e-02,\n",
              "        -6.610e-02,  4.890e-02,  4.210e-02,  5.200e-02,  2.200e-02,\n",
              "         1.156e-01,  3.700e-03, -6.560e-02, -1.450e-02, -5.720e-02,\n",
              "         1.290e-02, -4.400e-02,  4.180e-02, -0.000e+00, -6.390e-02,\n",
              "        -5.240e-02,  1.520e-02, -1.500e-02, -1.200e-02, -5.610e-02,\n",
              "         5.610e-02,  1.260e-02, -9.740e-02, -4.500e-02,  1.240e-02,\n",
              "        -3.260e-02, -1.440e-02, -2.520e-02, -1.210e-02,  2.610e-02,\n",
              "        -3.180e-02,  2.040e-02, -1.910e-02,  2.440e-02, -1.700e-02,\n",
              "         5.870e-02, -4.900e-03,  1.800e-03, -9.100e-03, -1.930e-02,\n",
              "        -8.110e-02,  3.580e-02,  1.300e-03,  8.300e-03,  1.355e-01,\n",
              "         1.370e-02, -1.027e-01, -3.870e-02,  9.200e-03,  7.200e-03,\n",
              "        -8.340e-02,  1.000e-04, -8.070e-02,  2.370e-02, -3.300e-02,\n",
              "        -5.600e-02,  5.090e-02, -2.200e-03,  3.640e-02, -7.500e-02,\n",
              "        -5.410e-02, -4.000e-02,  1.036e-01, -5.660e-02,  4.090e-02,\n",
              "        -5.570e-02,  5.640e-02, -6.030e-02,  3.950e-02,  2.450e-02,\n",
              "        -1.800e-03,  6.660e-02,  3.540e-02,  2.570e-02, -1.530e-02,\n",
              "        -2.570e-02,  3.060e-02,  6.840e-02, -5.610e-02, -2.910e-02,\n",
              "        -7.870e-02, -1.310e-02, -2.410e-02, -4.560e-02, -9.100e-03,\n",
              "         3.620e-02,  2.960e-02, -3.690e-02, -3.080e-02,  3.930e-02,\n",
              "        -7.760e-02, -1.540e-02,  1.210e-02, -1.369e-01, -1.600e-03,\n",
              "         5.870e-02, -1.437e-01, -2.900e-03, -3.290e-02, -1.290e-02,\n",
              "        -5.750e-02, -8.040e-02, -7.460e-02,  4.080e-02, -5.800e-03,\n",
              "        -1.490e-02,  3.210e-02,  8.440e-02,  4.290e-02, -2.510e-02,\n",
              "         3.260e-02, -2.510e-02,  1.390e-01, -3.700e-03, -1.359e-01,\n",
              "        -3.200e-02, -4.010e-02, -1.700e-03, -3.700e-02,  6.660e-02,\n",
              "        -2.690e-02, -3.800e-02, -7.120e-02, -7.820e-02, -5.090e-02,\n",
              "        -4.890e-02, -1.077e-01,  7.100e-03,  1.560e-02, -4.330e-02,\n",
              "         1.910e-02,  6.710e-02,  5.700e-02,  2.800e-03, -2.640e-02,\n",
              "         1.067e-01,  7.360e-02, -1.640e-02,  4.200e-02,  1.200e-02,\n",
              "        -5.260e-02,  1.308e-01,  1.910e-02,  6.660e-02, -6.340e-02,\n",
              "         2.550e-02,  2.950e-02, -1.800e-02, -7.660e-02,  3.130e-02,\n",
              "         9.300e-03,  1.100e-02, -2.990e-02, -6.930e-02,  1.300e-02,\n",
              "         4.610e-02,  7.930e-02, -7.920e-02,  3.000e-03, -4.240e-02,\n",
              "         1.720e-02,  2.500e-03, -7.250e-02,  6.380e-02, -2.350e-02,\n",
              "         2.800e-03,  4.090e-02,  1.460e-02, -1.750e-02,  3.970e-02,\n",
              "         8.940e-02, -3.230e-02,  5.700e-02,  8.600e-03,  7.900e-03,\n",
              "         1.020e-02, -7.820e-02, -1.072e-01, -1.063e-01,  6.660e-02,\n",
              "         8.980e-02, -7.500e-03,  9.310e-02, -7.260e-02, -2.620e-02,\n",
              "         1.740e-02,  5.700e-03, -7.850e-02, -8.430e-02, -3.740e-02],\n",
              "       dtype=float32),\n",
              " array([-0.0512,  0.0117,  0.0128,  0.0737,  0.093 ,  0.0535,  0.0127,\n",
              "        -0.0514,  0.0497, -0.0006,  0.0016, -0.0285,  0.0301,  0.085 ,\n",
              "         0.0518,  0.0465,  0.0103,  0.0103,  0.0301, -0.0077,  0.0567,\n",
              "         0.0149,  0.0393, -0.0087,  0.0109,  0.0811, -0.0628, -0.0578,\n",
              "        -0.038 , -0.0122, -0.0369,  0.0293,  0.0533, -0.0546, -0.0359,\n",
              "         0.0279, -0.0325, -0.0495, -0.0325,  0.0358, -0.0553, -0.0795,\n",
              "        -0.0129, -0.0317, -0.0614,  0.0756, -0.0378,  0.0329, -0.0998,\n",
              "        -0.0619,  0.0272, -0.022 ,  0.1001, -0.0874, -0.0029,  0.0525,\n",
              "        -0.1129,  0.0811,  0.0072,  0.0059,  0.0059, -0.0769,  0.0425,\n",
              "        -0.0491, -0.0811,  0.0051, -0.0723, -0.103 , -0.0513, -0.0999,\n",
              "         0.0405,  0.0217, -0.0311, -0.2088, -0.0215,  0.04  ,  0.0505,\n",
              "        -0.038 ,  0.0332, -0.0226, -0.0156,  0.0568, -0.0513, -0.0896,\n",
              "         0.1096, -0.0304,  0.0794,  0.0028, -0.1011,  0.0009,  0.0116,\n",
              "         0.0777, -0.1404, -0.0232,  0.0629, -0.0243,  0.0174,  0.0441,\n",
              "         0.0236, -0.0419, -0.0933,  0.0388, -0.047 ,  0.0309,  0.0178,\n",
              "         0.0639, -0.0349,  0.0346,  0.0683,  0.0722,  0.0544, -0.005 ,\n",
              "         0.1098,  0.0834, -0.0225, -0.0669, -0.0627, -0.0581, -0.0138,\n",
              "         0.0173,  0.0213, -0.0706,  0.0099, -0.016 ,  0.0414, -0.0616,\n",
              "         0.0019,  0.0319, -0.0082,  0.0805,  0.0465, -0.0624, -0.0078,\n",
              "         0.0085, -0.0005, -0.0659,  0.0744,  0.044 ,  0.0776, -0.0126,\n",
              "         0.0183,  0.016 , -0.0059, -0.0037,  0.0032, -0.0312, -0.005 ,\n",
              "         0.0321, -0.0536, -0.0701, -0.0471,  0.0549, -0.0083, -0.0515,\n",
              "         0.0993, -0.1592,  0.0233,  0.0216, -0.0319,  0.0093,  0.0145,\n",
              "        -0.0299, -0.0597,  0.0161, -0.0003,  0.0452,  0.1278,  0.0083,\n",
              "        -0.1115, -0.0094, -0.0507,  0.0746,  0.0652, -0.0299, -0.0752,\n",
              "        -0.1019,  0.0101,  0.0497,  0.0112,  0.0613,  0.0701, -0.1039,\n",
              "         0.0288, -0.0195, -0.0094, -0.053 , -0.0378,  0.0699, -0.0281,\n",
              "        -0.0129, -0.0314,  0.0098,  0.1019, -0.0331,  0.0692, -0.0209,\n",
              "         0.032 ,  0.0015, -0.0182, -0.081 ,  0.0697, -0.0931,  0.0702,\n",
              "        -0.0502,  0.0041,  0.086 , -0.0941, -0.0068, -0.026 , -0.0282,\n",
              "        -0.0618,  0.0525, -0.0343, -0.1023, -0.026 ,  0.0358,  0.0406,\n",
              "         0.1388,  0.0584,  0.0226, -0.1617, -0.0079,  0.0948, -0.1225,\n",
              "         0.0211, -0.0808, -0.0225,  0.0637, -0.0504, -0.0284,  0.003 ,\n",
              "         0.1072,  0.0081,  0.033 , -0.0778, -0.0159, -0.0689,  0.0007,\n",
              "        -0.048 ,  0.0245, -0.0738, -0.0805,  0.0121, -0.0064,  0.0613,\n",
              "        -0.0471,  0.0604, -0.0727,  0.0726, -0.0851,  0.024 ,  0.0262,\n",
              "        -0.1313, -0.0112, -0.0775,  0.0707,  0.0603, -0.0081,  0.0381,\n",
              "         0.0827,  0.0258, -0.0324, -0.0046, -0.0634, -0.0611, -0.0202,\n",
              "         0.0525, -0.0949, -0.0105,  0.027 ,  0.0193,  0.0425,  0.0745,\n",
              "        -0.0192,  0.1063,  0.076 ,  0.0565,  0.0484, -0.033 , -0.0078,\n",
              "         0.1517,  0.0261, -0.0409, -0.0463, -0.0457,  0.0315,  0.0273,\n",
              "        -0.0325, -0.0522,  0.0441,  0.0873, -0.0325,  0.0614,  0.0665,\n",
              "        -0.0068,  0.0222, -0.0298, -0.0999, -0.0526, -0.0085],\n",
              "       dtype=float32)]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 41
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a6kkyAYzUOgk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}